Sugu Sougoumarane

960 posts

Sugu Sougoumarane

Sugu Sougoumarane

@ssougou

Head of Multigres @Supabase, Co-creator @vitessio

Katılım Şubat 2012
164 Takip Edilen3.3K Takipçiler
Sabitlenmiş Tweet
Sugu Sougoumarane
Sugu Sougoumarane@ssougou·
I'm re-releasing the generalized consensus series one blog at a time for easier ingestion. Today, it will be the first two parts: Defining the Problem and Setting the Requirements. Links in 🧵
Sugu Sougoumarane tweet media
English
2
4
28
3.2K
Sugu Sougoumarane retweetledi
Qian Li
Qian Li@qianl_cs·
Our next South Bay Systems meetup will be on March 31. We've got two awesome deep-dive talks: - @ssougou will present deconstructing consensus and its application to Multigres - @stuhood will talk about how modern full-text search is evolving toward columnar systems (and the tricky optimization challenges) Food and beverages will be provided, courtesy of our hosts, @Snowflake Come hang out with the systems crowd 👇
Qian Li tweet media
English
1
5
10
779
Sugu Sougoumarane retweetledi
Supabase
Supabase@supabase·
We are pleased to inform our community that access to Supabase’s website supabase.co has now been fully restored for all users across India. We sincerely thank the Ministry of Electronics and Information Technology (MeitY) for their prompt action and constructive engagement in resolving this matter. Access to our services has been fully restored, and we deeply appreciate the continued patience and support of builders, developers, and businesses across India. Thank you for your patience while we worked through this incident. status.supabase.com
English
132
106
1.2K
152.1K
Sugu Sougoumarane retweetledi
Paul Copplestone - e/postgres
Paul Copplestone - e/postgres@kiwicopple·
Yesterday we had an outage affecting all services in the us-east-2 (Ohio) region We have published the post-mortem: supabase.com/blog/supabase-… This post is an account of what happened, how it happened, and the concrete steps we are taking to make sure it does not happen again. We know you depend on Supabase to be reliable. We let you down in both reliability and communications. I am grateful to every customer who reported issues, waited patiently, and provided feedback on how we can improve. We will make major structural changes to earn back your trust.
Paul Copplestone - e/postgres@kiwicopple

Today we had an outage that ran for 3h42m, affecting 4.92% of our customers. All systems in us-east-2 were affected. I'm sorry to everyone affected. There is no good excuse - you trust us with your infra and we need to do better. We already have mitigations in place and we're working on a post-mortem. We will post it in the next 12 hours.

English
50
34
284
123.6K
Sugu Sougoumarane retweetledi
Paul Copplestone - e/postgres
@coreyward Both Multigres & OrioleDB are landing this year and they will be industry-leading for performance
English
0
4
56
3.1K
Sugu Sougoumarane retweetledi
Paul Copplestone - e/postgres
we are looking for someone to lead CLI development at @supabase this is becoming one of the most critical interfaces for devs/agents tag the best Go dev that you know and if we hire them I'll give you $1,000 of Supabase credits They don't need CLI experience, they just need to have a good product sense. My pitch for Supabase: ◆ we're open source ◆ we don't do geo-adjusted pay ◆ we have an async work culture ◆ most engineers have 1 meeting per week ◆ we're full distributed across 50 countries ◆ we power the world's fasted growing companies Application link ↓ jobs.ashbyhq.com/supabase/1e7a9…
English
66
70
651
83.6K
Sugu Sougoumarane
Sugu Sougoumarane@ssougou·
@MarcJBrooker I almost replied yes also, but changed it to no: R+W formula works only for the positive case of discovering completed transactions. But you can also discover those that may later be abandoned due to failover: x.com/ssougou/status…
Sugu Sougoumarane@ssougou

@kellabyte Simple answer is no: You don't know if what you're reading has reached quorum. Any data that hasn't reached quorum can be rolled back if there's a failure. I've listed a few viable approaches here: multigres.com/blog/generaliz….

English
2
0
0
443
Sugu Sougoumarane
Sugu Sougoumarane@ssougou·
@kellabyte Simple answer is no: You don't know if what you're reading has reached quorum. Any data that hasn't reached quorum can be rolled back if there's a failure. I've listed a few viable approaches here: multigres.com/blog/generaliz….
English
0
0
1
611
Kelly Sommers
Kelly Sommers@kellabyte·
If you have consensus writes with agreement across all 3 nodes, a quorum (2 nodes) read would allow a slightly faster read path not waiting for the slowest node to respond while still guaranteeing read your own write semantics. Am I understanding this correctly?
English
9
0
13
8.6K
Joran Dirk Greef
Joran Dirk Greef@jorandirkgreef·
DistSys Interview Challenge Many factors may limit a system’s ability to scale. But which limit dominates a system’s ability to scale?
English
38
5
89
33.9K
Sugu Sougoumarane retweetledi
Paul Copplestone - e/postgres
@aakashgupta > They’ll recommend Supabase today and something better tomorrow Ironically the way we got to this point was to make it incredibly easy for our users to migrate away This Principle has been in our docs for 5 years now:
Paul Copplestone - e/postgres tweet media
English
11
15
153
16.3K
Sugu Sougoumarane retweetledi
Paul Copplestone - e/postgres
@multigres will land on supabase early next year features: ◆ Sharding ◆ Local NVMe disks ◆ Cross-zone failover ◆ Automated Backups ◆ Built-in Observability ◆ Change Data Capture ◆ Built-in Connection Pooler ◆ Built-in Database Protection ◆ Migrations across PG versions
Paul Copplestone - e/postgres tweet media
English
10
19
186
17K
Sugu Sougoumarane retweetledi
Sam Rome
Sam Rome@SamuelRome97104·
Supabase Culture: - work a lot on what you like - pants not needed - weekly updates - some deadlines - memes > - global slack channels - Supabase swag - permissionless - community - more memes - ant memes - launch weeks 🔴🟡🟢
English
8
8
106
13.3K
Sugu Sougoumarane retweetledi
Paul Copplestone - e/postgres
at @supabase we're building: ◆ a new storage engine for Postgres ◆ an open source sharding engine for Postgres ◆ an open source ETL engine for Postgres ◆ an FDW extension to query anything from Postgres we hire talented people who just want to build. come us ↓
English
55
60
989
95.3K
Ovais Tariq
Ovais Tariq@ovaistariq·
I am here at KubeCon Atlanta booth no. 1762, in case you want to chat about fast storage!
Ovais Tariq tweet media
English
3
2
30
2.2K
Sugu Sougoumarane
Sugu Sougoumarane@ssougou·
I'll be at Kubecon next week. Who else is going? Looking forward to meeting a lot of you.
Sugu Sougoumarane tweet media
English
1
1
7
419
Sugu Sougoumarane
Sugu Sougoumarane@ssougou·
@Eloq_Data Are you talking about a system like Aurora? I thought they avoid the need for checkpointing by directly materializing pages from the logs.
English
2
0
0
30
EloqData
EloqData@Eloq_Data·
@ssougou Fantastic read. In your proposed design for a generalized Raft, how do you see it impacting log management in a decoupled storage architecture? Curious if it simplifies the trade-offs between WAL placement and checkpointing to object storage.
English
1
0
2
37
Sugu Sougoumarane
Sugu Sougoumarane@ssougou·
The conclusion of the Generalized Consensus series is a recap of everything we covered. As a bonus, we propose a design to change Raft into a generalized solution. multigres.com/blog/generaliz…
Sugu Sougoumarane tweet media
English
2
2
13
674