Anthropic explains why it replaced Git with Scion to handle growing traffic from coding agents and CI jobs. Scion handles commits, merges, and history queries on the server, while clients fetch only the data they need through shared caches. The team reports checkouts about 100 times faster with much lower CPU and memory use. The design targets huge monorepos with unusually heavy traffic.
An engineering writeup on Pando, the Git compatible backend that serves OpenAI's monorepo. It explains how unrelated branches can accept pushes in parallel, while stored packfiles and regional caches make fetches fast. The key tradeoff is giving up one exact view of every branch so reads and writes can scale on their own.
A post from Cursor about Continuity, the Git storage system they built to host repositories at real scale. Instead of treating each repo as one precious copy on one disk, every push goes to a write ahead log in S3 first, and the client only hears back once that write is durable. Local disks become warm caches that can be rebuilt at any time, and replicas stay in sync with gossip over UDP plus conditional reads against S3. The post also covers repacking without making every replica redo the same expensive work, and gives real numbers: about 120 pushes a second on standard S3 and over 300 on S3 Express One Zone.