Full Width [alt+shift+f] Shortcuts [alt+shift+k]
Sign Up [alt+shift+s] Log In [alt+shift+l]

Stay updated

Get a weekly newsletter with the top 5 articles worth reading every week.

More from Jaz's Blog

Reverse Twins
2nd Oct 2025 24 votes
Turning Billions of Strings into Integers Every Second Without Collisions

I’ve recently started building a POC of a Redis RESP3 Wire Compatible Key/Value Database built on FoundationDB with @calabro.io and though it’s rather early, it’s already spawned a fun distributed systems problem that I thought would be interesting to share. Previously I’ve written about how I implemented a Graph DB via Roaring Bitmaps, representing relations as a bidirectional pair of sets. To support such use-cases in this new database, we’d like to represent sets of keys such that you can perform boolean operations on them (intersection, union, difference) relatively quickly even for very large sets (with millions of members). Supporting Larger Keys In the original Graph DB, we were representing user DID strings as uint32 UIDs to allow us to store millions of edge lists in very little space (e.g. the set of users who follow bsky.app) while being able to perform boolean operations between lists quickly (using Roaring Bitmaps’ parallel boolean operators). Since we were graphing follows, blocks, and other such User-to-User relationships, there was a practical maximum for the total number of user IDs in the low billions. We’ve continued exploring objects and relationships we’d like to represent as a Graph, and have realized that if we wanted to store e.g. the URIs of all posts a user has liked so we can intersect it with other users’ likes, we’re going to need a bigger keyspace! There are well over 15 Billion records in the AT Proto Ecosystem, each with a unique AT URI! Now our desired keyspace is much larger than can be represented by uint32 values and so we need to expand to uint64. Easy enough, let’s use the uint64 flavor of Roaring Bitmaps and simply intern URIs and User DIDs as uint64s, problem solved, right? Not quite… Interning Many Things at Once The AT Proto Firehose has hit historic peak traffic of over 1,500 evt/sec. We want to design a system that will handle many times more scale than we’ve ever seen in reality. This means designing for 10x or 100x would require us to be able to intern 15k to 150k new URIs per second into uint64 integers. Sounds easy enough, what’s the holdup? Well, in FoundationDB we’re able to use Transactions to do things like atomically increment a sequence safely when many other threads may be trying to do the same thing. This is simple enough to do in Go, we can just toss together a little helper function to acquire a new UID for our string: func (s *server) allocateNewUID(span trace.Span, tx fdb.Transaction) (uint64, error) { var newUID uint64 val, err := tx.Get(fdb.Key("last_uid")).Get() if err != nil { return 0, return fmt.Errorf("failed to get last UID: %w", err) } if len(val) == 0 { newUID = 1 // start from 1 } else { lastUID, err := strconv.ParseUint(string(val), 10, 64) if err != nil { return 0, return fmt.Errorf("failed to parse last UID: %w", err) } newUID = lastUID + 1 } tx.Set(fdb.Key("last_uid"), []byte(strconv.FormatUint(newUID, 10))) return newUID, nil } This function gets called from a fdb.Transaction which gets assigned a Transaction ID, then stages its changes, then tries to commit them. In FoundationDB, if your transaction is reading or modifying data written to by a different Transaction that finishes while you’re in-progress, your Transaction is thrown out and must be retried. For our UID assignment use-case, this is pretty problematic. We want to assign hundreds of thousands of new UIDs per second but if they’re all modifying the same key, concurrent transactions will constantly run into contention on the same data and will be forced to retry over and over again. This problem gets worse the more concurrent transactions you have trying to read from or write to the same key. Even if we stick to sequential access, if it takes ~5-10ms to assign a UID, we can only assign ~100-200 UIDs per second, nowhere near the throughput we need to support. How can we get past this problem and allow us to give strings unique uint64 UIDs in a high throughput and highly concurrent manner? Attempt #1: xxHash My first attempt to solve this problem was to try something that required no coordination and hash the string keys into uint64s using xxHash. xxHash is a non-cryptographic hash algorithm that supports incredibly high throughput (dozens of GB/sec) and can produce 64 bit unsigned integer hashes of strings trivially. Implementing this would look something like: Hash the incoming string key Lookup the uint64 UID to see if we’ve already assigned it to a string Reject the transaction if there’s a collision and give up Store the key in the UID map and the UID in the key map Use the UID for anything else we need While the uint64 keyspace is plenty large for our needs assuming we distribute evenly among the whole space, using a hashing algorithm with no coordination means there’s room for collisions and thus we’d need some additional logic (potentially by bucketing the keys somehow). Consulting the Birthday Problem we can see that a keyspace with 64 bit hashes has a >50% chance of containing a single collision when we have only ~5 billion keys in the set! That’s barely more keys than we can cram into a uint32 and definitely won’t suffice for the number of keys we expect to be storing! So, xxHash, while nice and coordination-free is probably not going to be the solution we need. What else can we do? Attempt #2: Billions of Sequences Incrementing one sequence is clearly not an option because we can only increment a single sequence ~100-200 times per second, but what if we instead had more than one sequence? Roaring Bitmaps managed to make highly efficient bitmap representations by breaking up a uint32 keyspace into a uint16-wide set of uint16-wide keyspaces. Can we do something similar here? Here’s an idea, what if we had just over 4 billion difference sequences and just picked one at random when we needed to assign a UID? Since we’re constructing our UIDs as a uint64, we can split the full UID into a pair of uint32s where the most-significant-bits are used to identify the sequence ID and the least-significant-bits are used to identify the value assigned to the UID within the sequence. So in our implementation, we get ~4.3 Billion sequence IDs that each have ~4.3 Billion incrementing values. As an example, if we were to randomly select Sequence ID 37 and then we increment that sequence to the value 5, we’d assemble the ID as 37<<32 + 5 which looks like 158,913,789,952 + 5 -> 158,913,789,957. Looking at the next Seuqence ID, we’d see 38 which, when left shifted by 32 gives us 163,208,757,248. You can see there’s a gap of ~4.3 billion values between the first UID assigned by each Sequence ID. Assuming we can increment a single sequence ~100 times per second with contention, we’re able to mint 430 Billion new UIDs per second without locking up (assuming the cluster can keep up). Storing ~4.3 billion sequences may be a bit expensive, but thankfully this strategy can scale up and down by picking a larger or smaller prefix size. If we only wanted to store say, ~16k sequences, we can pick a 14 bit prefix instead of a 32 bit prefix and then use a 50 bit sequence number. That spreads the load across 2^14 sequence IDs and significantly reduces storage requirements for Sequences. What does this look like in code? Well, it’s honestly not very complex! const uidSequencePrefix = "uid_sequence/" func (s *server) allocateNewUID(tx fdb.Transaction) (uint64, error) { // sequenceNum is the random uint32 sequence we are using for this allocation var sequenceNum uint32 var sequenceKey string // assignedUID is the uint32 within the sequence we will assign var assignedUID uint32 // Try up to 5 times to find a sequence that is not exhausted for range 5 { // Pick a random uint32 as the sequence we will be using for this UID sequenceNum = rand.Uint32() sequenceKey = fmt.Sprintf("%s%d", uidSequencePrefix, sequenceNum) val, err := tx.Get(fdb.Key(sequenceKey)).Get() if err != nil { return 0, fmt.Errorf("failed to get last UID: %w", err) } if len(val) == 0 { assignedUID = 1 // Start each sequence at 1 } else { lastUID, err := strconv.ParseUint(string(val), 10, 32) if err != nil { return 0, fmt.Errorf("failed to parse last UID: %w", err) } // If we have exhausted this sequence, pick a new random sequence if lastUID >= 0xFFFFFFFF { continue } assignedUID = uint32(lastUID) + 1 } } // If we failed to find a sequence after 5 tries, return an error if assignedUID == 0 { return 0, fmt.Errorf("failed to allocate new UID after 5 attempts") } // Assemble the 64-bit UID from the sequence ID and assigned UID newUID := (uint64(sequenceNum) << 32) | uint64(assignedUID) // Store the assigned UID back to the sequence key for the next allocation tx.Set(fdb.Key(sequenceKey), []byte(strconv.FormatUint(uint64(assignedUID), 10))) // Return the full 64-bit UID return newUID, nil } And there we go! We can now intern billions of strings per second with little to no contention in a distributed system while completely avoiding collisions and making full use of our keyspace! Conclusion Often times when designing distributed systems, patterns and strategies you see in seemingly unrelated libraries can inspire an elegant solution to the problem at hand. In the case of distributed, high-throughput string interning, horizontal scaling can be achieved by breaking up one large keyspace that requires strict coordination into billions of smaller keyspaces that can be randomly load-balanced across. Both patterns used in this technique are present elsewhere: Breaking up a large keyspace into a bunch of smaller keyspaces is present in Roaring Bitmaps (among other systems) Letting randomness and large numbers spread out resource contention is present in many load balancing systems This is one of my favorite parts of growing as an engineer: the more systems and strategies you familiarize yourself with, the more material you have to draw from when designing something new. Personal News A bit of personal news for y’all if you made it this far. Today is my last day as a member of the Bluesky team! The past 2+ years building out Bluesky’s Infrastructure and Platform team and scaling Bluesky from 100,000 -> 40,000,000 users have been the most intense and rewarding years of my life. I don’t have the words to express how much I’ve valued my time on the team and how much I care for the people I’ve worked with in what feels like a decade of real time. I’ve got some new adventures ahead and am excited to be embarking on a new journey within the next month (still building large-scale infrastructure, don’t worry). I plan to continue being involved in the AT Proto Community and to contribute to some cool projects other folks on the Bluesky team are working on from the FOSS space (like KVDB). To the team, I wish you all the best and will dearly miss getting to work with you all every day, but nothing lasts forever and I will always cherish the time I got to spend building an incredible platform with incredible people. If you’re interested in joining a world-class team doing important work, check out Bluesky’s open job listings here. There should be a new role opening up for a seasoned Go Engineer on the Platform team soon!

26th Sep 2025 1 votes
When Imperfect Systems are Good, Actually: Bluesky's Lossy Timelines

Often when designing systems, we aim for perfection in things like consistency of data, availability, latency, and more. The hardest part of system design is that it’s difficult (if not impossible) to design systems that have perfect consistency, perfect availability, incredibly low latency, and incredibly high throughput, all at the same time. Instead, when we approach system design, it’s best to treat each of these properties as points on different axes that we balance to find the “right fit” for the application we’re supporting. I recently made some major tradeoffs in the design of Bluesky’s Following Feed/Timeline to improve the performance of writes at the cost of consistency in a way that doesn’t negatively affect users but reduced P99s by over 96%. Timeline Fanout When you make a post on Bluesky, your post is indexed by our systems and persisted to a database where we can fetch it to hydrate and serve in API responses. Additionally, a reference to your post is “fanned out” to your followers so they can see it in their Timelines. This process involves looking up all of your followers, then inserting a new row into each of their Timeline tables in reverse chronological order with a reference to your post. When a user loads their Timeline, we fetch a page of post references and then hydrate the posts/actors concurrently to quickly build an API response and let them see the latest content from people they follow. The Timelines table is sharded by user. This means each user gets their own Timeline partition, randomly distributed among shards of our horizontally scalable database (ScyllaDB), replicated across multiple shards for high availability. Timelines are regularly trimmed when written to, keeping them near a target length and dropping older post references to conserve space. Hot Shards in Your Area Bluesky currently has around 32 Million Users and our Timelines database is broken into hundreds of shards. To support millions of partitions on such a small number of shards, each user’s Timeline partition is colocated with tens of thousands of other users’ Timelines. Under normal circumstances with all users behaving well, this doesn’t present a problem as the work of an individual Timeline is small enough that a shard can handle the work of tens of thousands of them without being heavily taxed. Unfortunately, with a large number of users, some of them will do abnormal things like… well… following hundreds of thousands of other users. Generally, this can be dealt with via policy and moderation to prevent abusive users from causing outsized load on systems, but these processes take time and can be imperfect. When a user follows hundreds of thousands of others, their Timeline becomes hyperactive with writes and trimming occurring at massively elevated rates. This load slows down the individual operations to the user’s Timeline, which is fine for the bad behaving user, but causes problems to the tens of thousands of other users sharing a shard with them. We typically call this situation a “Hot Shard”: where some resident of a shard has “hot” data that is being written to or read from at much higher rates than others. Since the data on the shard is only replicated a few times, we can’t effectively leverage the horizontal scale of our database to process all this additional work. Instead, the “Hot Shard” ends up spending so much time doing work for a single partition that operations to the colocated partitions slow down as well. Stacking Latencies Returning to our Fanout process, let’s consider the case of Fanout for a user followed by 2,000,000 other users. Under normal circumstances, writing to a single Timeline takes an average of ~600 microseconds. If we sequentially write to the Timelines of our user’s followers, we’ll be sitting around for 20 minutes at best to Fanout this post. If instead we concurrently Fanout to 1,000 Timelines at once, we can complete this Fanout job in ~1.2 seconds. That sounds great, except it oversimplifies an important property of systems: tail latencies. The average latency of a write is ~600 microseconds, but some writes take much less time and some take much more. In fact, the P99 latency of writes to the Timelines cluster can be as high as 15 milliseconds! What does this mean for our Fanout? Well, if we concurrently write to 1,000 Timelines at once, statistically we’ll see 10 writes as slow as or slower than 15 milliseconds. In the case of timelines, each “page” of followers is 10,000 users large and each “page” must be fanned out before we fetch the next page. This means that our slowest writes will hold up the fetching and Fanout of the next page. How does this affect our expected Fanout time? Each “page” will have ~100 writes as slow as or slower than the P99 latency. If we get unlucky, they could all stack up on a single routine and end up slowing down a single page of Fanout to 1.5 seconds. In the worst case, for our 2,000,000 Follower celebrity, their post Fanout could end up taking as long as 5 minutes! That’s not even considering P99.9 and P99.99 latencies which could end up being >1 second, which could leave us waiting tens of minutes for our Fanout job. Now imagine how bad this would be for a user with 20,000,000+ Followers! So, how do we fix the problem? By embracing imperfection, of course! Lossy Timelines Imagine a user who follows hundreds of thousands of others. Their Timeline is being written to hundreds of times a second, moving so fast it would be humanly impossible to keep up with the entirety of their Timeline even if it was their full-time job. For a given user, there’s a threshold beyond which it is unreasonable for them to be able to keep up with their Timeline. Beyond this point, they likely consume content through various other feeds and do not primarily use their Following Feed. Additionally, beyond this point, it is reasonable for us to not necessarily have a perfect chronology of everything posted by the many thousands of users they follow, but provide enough content that the Timeline always has something new. Note in this case I’m using the term “reasonable” to loosely convey that as a social media service, there must be a limit to the amount of work we are expected to do for a single user. What if we introduce a mechanism to reduce the correctness of a Timeline such that there is a limit to the amount of work a single Timeline can place on a DB shard. We can assert a reasonable limit for the number of follows a user should have to have a healthy and active Timeline, then increase the “lossiness” of their Timeline the further past that limit they go. A loss_factor can be defined as min(reasonable_limit/num_follows, 1) and can be used to probabilistically drop writes to a Timeline to prevent hot shards. Just before writing a page in Fanout, we can generate a random float between 0 and 1, then compare it to the loss_factor of each user in the page. If the user’s loss_factor is smaller than the generated float, we filter the user out of the page and don’t write to their Timeline. Now, users all have the same number of “follows worth” of Fanout. For example with a reasonable_limit of 2,000, a user who follows 4,000 others will have a loss_factor of 0.5 meaning half the writes to their Timeline will get dropped. For a user following 8,000 others, their loss factor of 0.25 will drop 75% of writes to their Timeline. Thus, each user has a effective ceiling on the amount of Fanout work done for their Timeline. By specifying the limits of reasonable user behavior and embracing imperfection for users who go beyond it, we can continue to provide service that meets the expectations of users without sacrificing scalability of the system. Aside on Caching We write to Timelines at a rate of more than one million times a second during the busy parts of the day. Looking up the number of follows of a given user before fanning out to them would require more than one million additional reads per second to our primary database cluster. This additional load would not be well received by our database and the additional cost wouldn’t be worth the payoff for faster Timeline Fanout. Instead, we implemented an approach that caches high-follow accounts in a Redis sorted set, then each instance of our Fanout service loads an updated version of the set into memory every 30 seconds. This allows us to perform lookups of follow counts for high-follow accounts millions of times per second per Fanount service instance. By caching values which don’t need to be perfect to function correctly in this case, we can once again embrace imperfection in the system to improve performance and scalability without compromising the function of the service. Results We implemented Lossy Timelines a few weeks ago on our production systems and saw a dramatic reduction in hot shards on the Timelines database clusters. In fact, there now appear to be no hot shards in the cluster at all, and the P99 of a page of Fanout work has been reduced by over 90%. Additionally, with the reduction in write P99s, the P99 duration for a full post Fanout has been reduced by over 96%. Jobs that used to take 5-10 minutes for large accounts now take <10 seconds. Knowing where it’s okay to be imperfect lets you trade consistency for other desirable aspects of your systems and scale ever higher. There are plenty of other places for improvement in our Timelines architecture, but this step was a big one towards improving throughput and scalability of Bluesky’s Timelines. If you’re interested in these sorts of problems and would like to help us build the core data services that power Bluesky, check out this job listing. If you’re interested in other open positions at Bluesky, you can find them here.

19th Feb 2025 63 votes
Emoji Griddle
30th Oct 2024 35 votes
Jetstream: Shrinking the AT Proto Firehose by >99%

Bluesky recently saw a massive spike in activity in response to Brazil’s ban of Twitter. As a result, the AT Proto event firehose provided by Bluesky’s Relay at bsky.network has increased in volume by a huge amount. The average event rate during this surge increased by ~1,300%. Before this new surge in activity, the firehose would produce around 24 GB/day of traffic. After the surge, this volume jumped to over 232 GB/day! Keeping up with the full, verified firehose quickly became less practical on cheap cloud infrastructure with metered bandwidth. To help reduce the burden of operating bots, feed generators, labelers, and other non-verifying AT Proto services, I built Jetstream as an alternative, lightweight, filterable JSON firehose for AT Proto. How the Firehose Works The AT Proto firehose is a mechanism used to keep verified, fully synced copies of the repos of all users. Since repos are represented as Merkle Search Trees, each firehose event contains an update to the user’s MST which includes all the changed blocks (nodes in the path from the root to the modified leaf). The root of this path is signed by the repo owner, and a consumer can keep their copy of the repo’s MST up-to-date by applying the diff in the event. For a more in-depth explanation of how Merkle Trees are constructed, check out this explainer. Practically, this means that for every small JSON record added to a repo, we also send along some number of MST blocks (which are content-addressed hashes and thus very information-dense) that are mostly useful for consumers attempting to keep a fully synced, verified copy of the repo. You can think of this as the difference between cloning a git repo v.s. just grabbing the latest version of the files without the .git folder. In this case, the firehose effectively streams the diffs for the repository with commits, signatures, and metadata, which is inherently heavier than a point-in-time checkout of the repo. Because firehose events with repo updates are signed by the repo owner, they allow a consumer to process events from any operator without having to trust the messenger. This is the “Authenticated” part of the Authenticated Transfer (AT) Protocol and is crucial to the correct functioning of the network. That being said, of the hundreds of consumers of Bluesky’s production Relay, >90% of them are building feeds, bots, and other tools that don’t keep full copies of the entire network and don’t verify MST operations at all. For these consumers, all they actually process is the JSON records created, updated, and deleted in each event. If consumers already trust the provider to do validation on their end, they could get by with a much more lightweight data stream. How Jetstream Works Jetstream is a streaming service that consumes an AT Proto com.atproto.sync.subscribeRepos stream and converts it into lightweight, friendly JSON. If you want to try it out yourself, you can connect to my public Jetstream instance and view all posts on Bluesky in realtime: $ websocat "wss://jetstream2.us-east.bsky.network/subscribe?wantedCollections=app.bsky.feed.post" Note: the above instance is operated by Bluesky PBC and is free to use, more instances are listed in the official repo Readme Jetstream converts the CBOR-encoded MST blocks produced by the AT Proto firehose and translates them into JSON objects that are easier to interface with using standard tooling available in programming languages. Since Repo MSTs only contain records in their leaf nodes, this means Jetstream can drop all of the blocks in an event except for those of the leaf nodes, typically leaving only one block per event. In reality, this means that Jetstream’s JSON firehose is nearly 1/10 the size of the full protocol firehose for the same events, but lacks the verifiability and signatures included in the protocol-level firehose. Jetstream events end up looking something like: { "did": "did:plc:eygmaihciaxprqvxpfvl6flk", "time_us": 1725911162329308, "type": "com", "commit": { "rev": "3l3qo2vutsw2b", "type": "c", "collection": "app.bsky.feed.like", "rkey": "3l3qo2vuowo2b", "record": { "$type": "app.bsky.feed.like", "createdAt": "2024-09-09T19:46:02.102Z", "subject": { "cid": "bafyreidc6sydkkbchcyg62v77wbhzvb2mvytlmsychqgwf2xojjtirmzj4", "uri": "at://did:plc:wa7b35aakoll7hugkrjtf3xf/app.bsky.feed.post/3l3pte3p2e325" } }, "cid": "bafyreidwaivazkwu67xztlmuobx35hs2lnfh3kolmgfmucldvhd3sgzcqi" } } Each event lets you know the DID of the repo it applies to, when it was seen by Jetstream (a time-based cursor), and up to one updated repo record as serialized JSON. Check out this 10 second CPU profile of Jetstream serving 200k evt/sec to a local consumer: By dropping the MST and verification overhead by consuming from relay we trust, we’ve reduced the size of a firehose of all events on the network from 232 GB/day to ~41GB/day, but we can do better. Jetstream and zstd I recently read a great engineering blog from Discord about their use of zstd to compress websocket traffic to/from their Gateway service and client applications. Since Jetstream emits marshalled JSON through the websocket for developer-friendliness, I figured it might be a neat idea to see if we could get further bandwidth reduction by employing zstd to compress events we send to consumers. zstd has two basic operating modes, “simple” mode and “streaming” mode. Streaming Compression At first glance, streaming mode seems like it’d be a great fit. We’ve got a websocket connection with a consumer and streaming mode allows the compression to get more efficient over the lifetime of the connection. I went and implemented a streaming compression version of Jetstream where a consumer can request compression when connecting and will get zstd compressed JSON sent as binary messages over the socket instead of plaintext. Unfortunately, this had a massive impact on Jetstream’s server-side CPU utilization. We were effectively compressing every message once per consumer as part of their streaming session. This was not a scalable approach to offering compression on Jetstream. Additionally, Jetstream stores a buffer of the past 24 hours (configurable) of events on disk in PebbleDB to allow consumers to replay events before getting transitioned into live-tailing mode. Jetstream stores serialized JSON in the DB, so playback is just shuffling the bytes into the websocket without having to round-trip the data into a Go struct. When we layer in streaming compression, playback becomes significantly more expensive because we have to compress outgoing events on-the-fly for a consumer that’s catching up. In real numbers, this increased CPU usage of Jetstream by 23% while lowering the throughput of playback from ~200k evt/sec to ~28k evt/sec for a single local consumer. When in streaming mode, we can’t leverage the bytes we compress for one consumer and reuse them for another consumer because zstd’s streaming context window may not be in sync between the two consumers. They haven’t received exactly the same data in the session so the clients on the other end don’t have their state machines in the same state. Since streaming mode’s primary advantage is giving us eventually better efficiency as the encoder learns about the data, what if we just taught the encoder about the data at the start and compress each message statelessly? Dictionary Mode zstd offers a mechanism for initializing an encoder/decoder with pre-optimized settings by providing a dictionary trained on a sample of the data you’ll be encoding/decoding. Using this dictionary, zstd essentially uses it’s smallest encoded representations for the most frequently seen patterns in the sample data. In our case, where we’re compressing serialized JSON with a common event shape and lots of common property names, training a dictionary on a large number of real events should allow us to represent the common elements among messages in the smallest number of bytes. For take two of Jetstream with zstd, let’s to use a single encoder for the whole service that utilizes a custom dictionary trained on 100,000 real events. We can use this encoder to compress every event as we see it, before persisting and emitting it to consumers. Now we end up with two copies of every event, one that’s just serialized JSON, and one that’s statelessly compressed to zstd using our dictionary. Any consumers that want compression can have a copy of the dictionary on their end to initialize a decoder, then when we broadcast the shared compressed event, all consumers can read it without any state or context issues. This requires the consumers and server to have a pre-shared dictionary, which is a major drawback of this implementation but good enough for our purposes. That leaves the problem of event playback for compression-enabled clients. An easy solution here is to just store the compressed events as well! Since we’re only sticking the JSON records into our PebbleDB, the actual size of the 24 hour playback window is <8GB with sstable compression. If we store a copy of the JSON serialized event and a copy of the zstd compressed event, this will, at most, double our storage requirements. Then during playback, if the consumer requests compression, we can just shuffle bytes out of the compressed version of the DB into their socket instead of having to move it through a zstd encoder. Savings Running with a custom dictionary, I was able to get the average Jetstream event down from 482 bytes to just 211 bytes (~0.44 compression ratio). Jetstream allows us to live tail all posts on Bluesky as they’re posted for as little as ~850 MB/day, and we could keep up with all events moving through the firehose during the Brazil Twitter Exodus weekend for 18GB/day (down from 232GB/day). With this scheme, Jetstream is required to compress each event only once before persisting it to disk and emitting it to connected consumers. The CPU impact of these changes is significant in proportion to Jetstream’s incredibly light load but it’s a flat cost we pay once no matter how many consumers we have. (CPU profile from a 30 second pprof sample with 12 consumers live-tailing Jetstream) Additionally, with Jetstream’s shared buffer broadcast architecture, we keep memory allocations incredibly low and the cost per consumer on CPU and RAM is trivial. In the allocation profile below, more than 80% of the allocations are used to consume the full protocol firehose. The total resident memory of Jetstream sits below 16MB, 25% of which is actually consumed by the new zstd dictionary. To bring it all home, here’s a screenshot from the dashboard of my public Jetstream instance serving 12 consumers all with various filters and compression settings, running on a $5/mo OVH VPS. At our new baseline firehose activity, a consumer of the protocol-level firehose would require downloading ~3.16TB/mo to keep up. A Jetstream consumer getting all created, updated, and deleted records without compression enabled would require downloading ~400GB/mo to keep up. A Jetstream consumer that only cares about posts and has zstd compression enabled can get by on as little as ~25.5GB/mo, <99% of the full weight firehose. Feel free to join the conversation about Jetstream and zstd on Bluesky.

24th Sep 2024 41 votes

More in programming

Mommy bloggers react

After a write-up in the New York Times, Mommy Bloggers had two options. Either lean in, or step back. Given how popular it became after that, it's not hard to guess which option they chose. The post Mommy bloggers react appeared first on The History of the Web.

10 hours ago 1 votes
Haunt 0.4.0 released

I'm quite a bit late on this one, but Haunt version 0.4.0 was released released back in July. I haven't had much time for blogging, but I'm catching up now! This release contains a small set of improvements and bug fixes since the 0.3.0 release in 2024. About Haunt Haunt is a static site generator that uses the Guile Scheme as its configuration language. It aims to be simple, functional, and extensible. Features include: Easy blog and Atom/RSS feed generation Markdown post support Simple development server for viewing edits before publishing Purely functional build process User extensibility Notable changes Added support for HTML in Markdown documents. This was a long time coming because guile-markdown did not support it and the library was abandoned by the original maintainer. As part of my work at Spritely, we forked it, implemented the relevant portions of the CommonMark specification, and released it. Spritely's guile-commonmark fork is now considered to be the official upstream by Guix and others. A further consequence of this is that guile-lib is now a required dependency for building Haunt as we need the (htmlprag) module to parse Markdown documents with embedded HTML. html->shtml from guile-lib's (htmlprag) module is now used instead of xml->sxml in the HTML reader. It was silly of me to use xml->sxml for this purpose years ago, but at the time I wanted guile-lib to be an optional dependency. Added haunt new subcommand for creating a new site. Added default directory, template, and prefix arguments to flat-pages procedure. Added support for index metadata flag to flat pages for pretty URLs. Flat pages now receive all page metadata, not just the page title. This is a breaking change from 0.3.0. Added .scm as an additional extension for sxml-reader. make-file-extension-matcher now supports multiple extensions. Fixed emission of <script> and <style> elements. Fixed handling of no available reader in flat pages builder. Fixed unreachable error handling clause when a reader is not found for a post. Fixed default blog theme template missing an <html> tag. Fixed overloaded -h option in haunt serve. Deprecated post in Skribe reader in favor of document. Download Haunt 0.4.0 is already available in Guix: guix pull guix install haunt See the Haunt project page for information on how to build from source. Thank you to Camilo Rodrigues, Noé Lopez, jgart, Jakob L. Kreuze, and Daniel Meißner for their contributions to this release! Happy haunting!

12 hours ago 1 votes
On reading books

How books have coloured my life

14 hours ago 1 votes
I Bought A Scanner (No, Really This Time)

This is a transcript from a talk I gave at the German Perl Workshop earlier this year. If you'd prefer to watch the video recording, you can find it here. I have lots of photographic projects on the go. Lots of these being on film, as some of these I started shooting a long time ago. I don’t have any particular loyalty or attraction to film, it’s just that I started shooting many of these projects before affordable medium format digital was available. Since I mostly shoot medium/large format film I never really jumped to digital until recently, so film has continued to feature heavily in my workflow. That said, it’s a pain in the arse to shoot film now given the spiraling costs, limited availability, and issues around traveling with it: modern airport CT scanners, being rolled out across many airports, are much more convenient but will fog film. Asking for a hand inspection often comes down to arbitrary timing - how busy the security is, how experienced the operator is, or if you’re lucky/unlucky. I’ve had film forced to be scanned (and fogged) and politely argued with security on more than one occasion. I don’t want to deal with that so don’t travel with film anymore, thus I am shooting less of it and have mostly moved to digital. I still have a tonne of film I need to scan and process however. Here’s just some of the binders and files of film. I don’t plan to scan all of this, but I do plan to scan the ones I need to. Probably in the region of a couple of thousand frames. I want to scan to the highest possible quality (within reason) for archiving, book projects, and large prints. If you’re wondering how large I print, it can be up to 160x60cm panoramics for selling. This is restricted by the size of my printer (that’s another story). Three Years Ago Three years ago I almost bought a scanner. I ended up blogging about it and the post got a bit of traction on Hacker News (HN). I’m never quite sure which posts I submit will pique the interest of the users. I’ll spend months chipping away at a draft and when I post it it tanks. Or I’ll cobble something together in twenty minutes, like the linked one above, and it gets 440 points and over 300 comments… The thread had some useful suggestions and some not so useful ones, the not so useful ones being effectively “buy an Epson”: I’ve had one for fifteen years and it’s not good enough for large prints or archiving. It’s passable for web stuff and smaller prints, but for my recent use cases? Not even close. Ten years ago I had negatives scanned with a high resolution scanner for the first time and recently, wanting to scan my archives for various projects, I decided I should invest in one of those scanners. The Original Plan The plan, back in 2023, was simple: Buy scanner (at significantly reduced rate) Scan all my film Sell scanner Profit! And I mean profit - the scanner that I almost bought was being offered to me at about 2/3rd of the price they usually sell. And they’re becoming harder to find in working order so the prices are going up. Or profit in not having to pay > 25.- CHF per frame to have someone else do this. You can see the pricing from The Film Lab. You can read the original blog post to find out more about the scanner in question, so I won’t repeat it here. Other than the parts being relevant to the rest of this post, namely that the scanner was showing hard and soft problems. The software that drives the scanner was last updated in 2012, it’s proprietary and closed source, requiring 32bit architecture and no third party drivers or software exist. So you are stuck using old software/computers to run it. Or maybe you could use emulation / virtualisation? The problem there is that the interface is firewire, or SCSI on the even older models, and firewire is known to be problematic on these scanners as the controllers start to go bad after a decade of continued use. That’s a risk, and the scanner was very much EOL as the firewire controller was dying: both ports were bad that suggests controller, not ports. The scanner would have been €5,000 to purchase and then €3,000 (ish) to repair. Or, as HN suggested - just open it up and use a soldering iron. I’m not going to drop 5k on something and then start poking it with a soldering iron. I’ll pass on that thanks. Camera Scanning In the meantime I’ve been camera scanning, which you can read about in another blog post. But how does that compare cost wise? It’s expensive because you’ll need a high resolution camera, a macro lens, copy stand, negative carrier/holder, and quality light source. You’ll look to spend anything from three to five thousand Euros on everything. Camera scanning does actually work well, in that it’s close to a high resolution dedicated scanner. But you have to setup the entire thing every time you want to use it, including ensuring everything is straight and parallel. It also suffers from the same weakness as most other scanning methods. What do you think that is? Film Flatness Or lack thereof: Film is rarely flat, especially so with 35mm. These are pretty mild examples of curl. It tends to be flatter in the larger formats but then you get into flatness issues due to it sagging. The smallest difference in the film plane can cause major issues in sharpness due to focus fall off (film scanning is essentially macro photography). Any workflow or solution that does not take this into account is significantly compromised. And the workflow is only as good as its weakest part. This is the biggest problem in scanning film - all other considerations are more than adequate these days: resolution, dynamic range, etc. However, most negative carriers don’t keep the film perfectly flat. This has always been a problem - this is from a book called “Edge of Darkness” which is about traditional analog photography and printing, and summarises the problems of negative carriers thusly: “if you use a glassless negative carrier, you might as well just buy the cheapest enlarging lens you can find. You are simply throwing away the money and sharpness you paid for it in your enlarging lens, and also in your fine camera and the expensive lenses you bought for it… No film will lie flat in a glassless carrier. That’s right, none… There is no avoiding this issue. Use glass.” So you have to use (anti-newton ring) glass, which introduces other issues - you’ve now got extra glass in the transmission path, and dust (which isn’t a massive problem, but a pain nonetheless). You could use drum scanning, which is absurdly impractical from a cost and operating point of view. Or you could use a Flextight, the scanner I almost bought three years ago. Interim Solution I stuck with camera scanning, but wasn’t happy though, because of film flatness and the setup faff. So of course I started looking for another scanner. I was idly browsing near the end of 2025 and came across this one. It’s exactly the same spec as the one I tried three years ago, except SCSI not Firewire so less prone to failure. It just predates Hasselblad buying Imacon (so is pre the rebranding, etc). It was in Switzerland so I could inspect and pick it up. It was also significantly cheaper than the previous one I had looked at, so worth a punt even if I needed to take a soldering iron to it. We went to St Gallen for a weekend and I picked it up. Here’s the software interface back in my studio. Look at that marvelous interface! None of that liquid glass bollocks. The first scans were promising, but I had the sense things needed some TLC. The first thing was calibrating the focus, which the software can do in combination with a focus slide. I was lucky that the focus slide was included with the scanner and I’m not sure what I would have done otherwise. Probably paid a fortune for a replacement? Possibly a lot of manual trial and error with the software? After doing that I scanned images of the 1951 USAF resolution test chart (taken on ultra high resolution 35mm film): That’s what the resulting scan looked like. Notice that it’s sharp from edge to edge, corner to corner. At 100% crop we can resolve around 110 to 123 line pairs per mm, which equates to about 5,600 to 6,300 DPI. This is beyond the limit of most 35mm lenses, but importantly - exactly to spec for this scanner. So I was happy the focus was calibrated. If you’re curious this is the same target with the camera scanning setup. It’s close, but we’ve got another variable in the workflow, several even, and that impacts the results. It’s not as sharp, and the extra glass in the transmission path causes aberrations. Another thing that needed attention was the power supply. The seller mentioned that “sometimes it takes five minutes to warm up”. Sometimes it was more than five minutes, and the power supply would click click click away. So that needed fixing and it was easy enough to find a compatible new replacement, however it cost 200 Euros. Expensive! The third problem I noticed was that some of the scans were coming out stretched. Often about 10% too wide/long, sometimes more than that. My panoramics looked panoooooooramic. I did some research and someone suggested this might be a “buffering issue”, which I thought was nonsense. Doing some testing I heard slipping sounds when the scanner was pulling the film into the body. After more research I stumbled on a post that suggested the belts need replacing. I opened the scanner up, and sure enough: A ha! You can’t quite see that the one on the back is even worse. I replaced those with compatible belts: 535 synchroflex t 2.5/245. Problem solved. The fourth problem was that the film holders were old and/or had been mishandled. They were falling apart and held together with electrical tape or glue, which didn’t seem optimal. Replacements cost 350 Euros in total for the four I needed. They’re now available cheaper from China, since the patents have expired. Or, you know, China. They used to cost about 200 Euros each from Hasselblad. The fifth problem, which is a potential one and hasn’t manifested yet, is that the lamps may eventually need replacing. I picked up a couple for 25 Euros. That seemed like a reasonable thing to do while they’re still available. Success? Let’s add up the costs of acquiring this scanner and renovating it: Scanner: 1,750.- CHF Power Supply: 175.- CHF Belts: 25.- CHF Film Holders: 350.- CHF Lamps: 25.- CHF Total: 2,325.- CHF (c. 2,500 EUR) In the last year (since acquiring the scanner) I have scanned: c. 250 panoramics frames (~ 6,000 CHF) c. 2,500 medium format frames (~ 80,000 CHF) c. 200 large format frames (~ 9,000 CHF) The figures in parentheses are what it would have cost me to have that number of frames scanned by a third party. That is, er, quite a saving. Also quite a lucrative business model perhaps? I think I can argue the cost of the scanner was a very good investment, and I haven’t finished using it yet. Even if it were to stop working tomorrow, it has already paid for itself many times over. Could it stop working tomorrow? Yes, because of other issues that will be harder to solve. The Bigger Issue(s)? A Power Mac G4 (discontinued in 2004). This came with the scanner, the necessary hardware and software to drive it, and is almost certainly living on borrowed time. Spinning metal is never good in the long-term. I’ll maybe purchase a backup soon, as these can still be found for a couple of hundred Euros. The key thing though, is that this very expensive, very high quality scanner, will at some point be rendered useless by the upgrade treadmill because the software required to run it will be increasingly difficult to run. A scanner that is still used by businesses, educational institutions, and individuals like me. A scanner that originally cost tens of thousands of Euros less than a decade ago. The upgrade treadmill is constantly whirring away. This is from the top of the Seattle Space Needle. “Do not upgrade anything on computer”. Clearly that notice speaks of someone being bitten by an upgrade at some point. I wonder is anyone else feeling the fatigue? Security updates, sure I can understand. But feature creep and trivialities? No! What tangible benefits have the last ten, fifteen, or even twenty years of OS updates brought? Other than security, and compatibility with newer hardware? New hardware is great, really, but by association forced deprecation of older hardware. No! It feels like the upgrade treadmill gets faster and steeper every year. Add to that subscription lock-in and dead endpoints: “I couldn’t vacuum my house because an SSL cert had expired” is what someone told me earlier this year. Fortunately this person is a software engineer so ended up man-in-the-middling the network traffic to get the vacuum cleaner to work again (no SSL-pinning it seems). “GoPro is announcing the end of life of the GoPro Quik app for macOS, effective at the end of 2024”. They discontinued the former in favour of their mobile app, which requires an account, login, subscription, and so on. I just want to transfer the videos from the hardware, I don’t need any of this crap (I don’t need any of that crap, it turns out GoPro haven’t locked the device down enough to prevent using third party apps to access the files. Yet). And, of course, software has to be in everything. These days the scanner would/could have an embedded Raspberry PI? Just a keyboard and mouse input, monitor and USB output would reduce the surface area, connectivity issues, and software dependency. Or software is never done? Because: externalities. I guess software is “done” when it’s no longer supported? Marciano Planque has a good piece on this: When hardware products reach end-of-life (EOL), companies should be forced to open-source the software. I think that’s a fair thing to say. I suspect Hasselblad/Imacon never open-sourced the software due to licensing issues. Or they just lost the source. Or they just don’t care, I don’t know. Maybe some combination of the three. And, inevitably, discontinued hardware like this scanner. Or, that is to say, discontinued parts? What about regulation changes? The panoramics I shoot are with a camera that was discontinued in 2004 because EU regulation banned lead solder in circuit boards. The company decided redesigning the parts wasn’t worth it. Old hardware has new exciting ways to fail. As time goes on components will fail or loosen - components that were expected to last decades. Then that results in tribal knowledge, or worse link rot and QR code rot. A lot of this stuff is hidden in walled gardens. There’s a Facebook Imacon group, for example. Why in the ever-loving fuck is a group for technical people, by technical people, on Facebook? Then there’s misleading AI. “My flextight scans are coming out stretched, what might the problem be?” LLM’s have gobbled up all the right information, and all the wrong information. Or information that is massively out of date. Nowhere in the suggestions here does it mention the belts might need replacing, which, according to my own research, is the most common reason these days. Legacy Software A decade ago I wrote an essay that also hit the front page of HN: All Software is Legacy. I think it is still relevant today, some parts not so much given we are now in The Age of Prompt, but mostly it’s still true. Nicholas always said “legacy software is the ugly stuff that makes you money”, which I think is true. But now it’s the stuff that surrounds us, like when I want to withdraw cash (guess what software most cash machines are still running?). Or when I want to take a train - when I gave this talk in Germany I had to get from the airport to the city centre. The ticket machines were disabled with a sign saying “no longer in use, download the app”. Then register. Then buy the ticket. I just want to give you money. Or when I wanted to pay for parking while stopping off at some random town in the UK - the same situation as with the ticket machines. “Download the app, register, pay”. Fuck that, I went and parked somewhere else. I just want to park, I don’t want to fight with software. Or if I want to hire a bike (not pictured: the half dozen apps on my phone to hire a bike). And when I want to buy stuff from a shop… One of the self-checkouts crashed recently in the coop, rebooting into a version of SUSE Linux from well over a decade ago. We’re collectively creating more and more of this everyday, letting it out into the world where it becomes a future liability for someone or the death knell for something. A pile of bikes, an unplugged ticket machine, a top of the line but no longer driveable scanner. References Imacon Users Group (the non-Facebook group) The state of Hasselblad Flextight scanners (2019) 1951 USAF resolution test chart Vlads Test Target Printer Story Original Scanner Blog Responses to HN Camera Scanning All Software is Legacy Repair Cafe

yesterday 1 votes
Attention is all you have

The Tetris effect is one of psychology’s most easy to reproduce experiments. Simply spend a bit of time playing the eponymous game every day for a few weeks. After a little while, you’ll start recognizing familiar Tetromino shapes in clouds, buildings, and everyday objects. You might even see them appear before your eyes when you start falling asleep. Tom Tang Attention hijacking There’s one lesson the Tetris effect teaches us: whatever you focus on long enough will end up shaping your thoughts. This can be a good thing since it’s how we learn new skills and discover new ideas. Sadly, less and less of our attention is focused intentionally. Instead of picking what we want to see we let other people decide what is supposed to be good for us. Do you want to watch a video? YouTube knows you like cooking and art streams. But why not also recommend a few clips about the stock market bubble, global warming, and the war in Iran. Doomscrolling will make you stay longer and click on a few more ads. Do you want to listen to music? Just open a Spotify playlist and let the algorithm figure out what you like. Please ignore the AI slop they will insert in between real songs to avoid paying royalties to real artists. Do you want to know how your colleagues are doing? Too bad, LinkedIn will bury any relevant career news between the opinion of complete strangers. It is surely just a coincidence that those strangers happen to be shilling whatever Microsoft is invested in at the moment. Do you want the opinion of strangers on a product? Well those Redditors you wanted to ask are probably just a bunch of LLMs talking to a bunch of Russian trolls now. I hope you didn’t value their opinion too much. If, like me and most people, you spend the major part of your day focused on your device, there’s no doubt it’s affecting you. And when you let someone else dictate what appears on your screen, it’s the same as giving them the key to your brain. New York Said Back to an intentional internet The internet wasn’t always like that. Before recommendation algorithms where a thing, you had to decide what you would be doing on the computer. You didn’t really have one big app that you could open and order it to entertain you. Instead, you had a few dozen of bookmarks to websites, each with a specific idea in mind. A site for video game news, that one website with lots of tutorials, a blog about anime that didn’t update often enough, a wiki about a TV show from the 90s… Of course awful things existed on the web. We had Encyclopedia Dramatica and Rotten.com, but you actually had to put the effort to go there if you wanted. Nobody was going to put pictures of dead kids and far-right propaganda as a suggestion after a pancake recipe or a cat video. The good thing is that this intentional internet is still around. It has just been a bit buried below the corporate web, but it’s not very hard to find. After all you’re on this blog, so you probably already have a good idea about it. The main difference between this time and now is you. When you want to get back to reading blogs, RSS feeds, and finish that tutorial instead of doomscrolling shorts, you have to get used to a slower internet. One where content is not infinite and doesn’t get updated every click. But like every habit, the only thing you have to do is to keep at it. And if you pay enough attention to it, something will click in your brain.

2 days ago 2 votes
📚 BoredReading

You seem to be enjoying this.

Join free to unlock everything.

Create free account

Already have an account? Sign in