From 07c45518716176e85f1b91937088c33b27cbecaf Mon Sep 17 00:00:00 2001 From: Ralf Kistner Date: Tue, 29 Oct 2024 17:26:41 +0200 Subject: [PATCH] Update limits on synced rows. --- resources/performance-and-limits.mdx | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/resources/performance-and-limits.mdx b/resources/performance-and-limits.mdx index 37cf8cd9..a2d3ebc3 100644 --- a/resources/performance-and-limits.mdx +++ b/resources/performance-and-limits.mdx @@ -9,7 +9,7 @@ The PowerSync Cloud Enterprise plan allows for limits to be customized. * Number of synced buckets per user: 1,000. * Sync requests where this number is exceeded will fail with a hard error. -* Number of rows synced per client: Currently scales efficiently to around 10,000 - 100,000 rows, with plans to scale to over 1M rows per client soon. + * We have plans to increase this limit. * Maximum row size: 15MB * This applies to both the source Postgres row, and the transformed row synced to the client. * Number of concurrent connections per [PowerSync Service](/architecture/powersync-service) instance: Limited to 3,000 by default. @@ -17,7 +17,10 @@ The PowerSync Cloud Enterprise plan allows for limits to be customized. * Expect a peak of around 2,000-4,000 operations/second for small rows, or 5MB/second for large rows. * This also applies to reprocessing sync rules or adding new tables. * Smaller transactions are processed at around 60 transactions/second. -* Synced rows (PowerSync Service -> Client): Expect around 1,000-10,000 operations/second/client, depending on the client. +* Synced rows (PowerSync Service -> Client): + * Over 1M rows per client is supported, with no hard limit. + * Expect a rate of around 2,000-20,000 operations/second/client, depending on the client. + * Database size and initial sync time may impose practical limits on number of rows. * Storage size: A soft limit of 100GB for data stored on the PowerSync Service by default. * Number of unique users: No hard limit. * Number of tables: No hard limit, but having hundreds of tables may slow down startup and sync performance.