PgConf.Russia 2020

Open-source maintainers face many challenges as projects grow. How to write more required features, fix more issues and have time to watch more pull requests? On the example of WAL-G(backup tool for PostgreSQL) I will tell you about how we solved these problems by launching a course of Open-source development at Ural Federal University, what we achieved and what will we do next.

Last year I made a talk about unexpected PostgreSQL bottlenecks, which could make sad surprise to user (or DBA). Feedback to my talk was very positive. Additionally I have new material after year. This is why I'm making a sequel including new unexpected situations when your database hangs. This time focus will be on multicore hardware platforms, but not only them.

A brief story how MySQL → PG migration could increase company efficiency tenfold times:

1. Program code has been reduced 50 times, with optimization of backend team (from 15 to 3 engineers)
2. Software development of new features has become measuring in days, not in months
3. Infrastructure costs per 1M users have been reduced 20 times
4. Database structure and technical documentation were simplified significantly, from 100K high-dependent tables to just 20 simple tables
5. New security level because of total forbidden on external SQL commands to the database
6. Quick analytics aggregation on multiple parameters, without external analytics systems
7. The last, but not the least: the main business was keeping alive during migration

Cloud storage has some unique characteristics compared to traditional storage mainly because it is virtualized and controlled by software. One example is that AWS EBS shows higher throughput with larger I/O size up to 256 KiB without hurting latency. Hence, a user can get only about 4 MiB/sec with 1,000 IOPS EBS volume if the I/O request size is 4 KiB, whereas a user can get about 250 MiB/sec if the I/O request size is 256 KiB. This is because EBS consumes one I/O in a given IOPS budget for every I/O request regardless of the I/O size (up to 256 KiB). Unfortunately, PostgreSQL cannot exploit the full potential of cloud storage because PostgreSQL has designed without considering the unique characteristics of cloud storage.

In this talk, I will introduce the AppOS extension that improves the throughput of a write-intensive workload by 10x by transparently making PostgreSQL cloud storage-native. AppOS works like a storage driver that efficiently exploits the characteristics of cloud storage, such as I/O size dependency to storage throughput and latency, atomic write support in cloud block storage, and fast, but non-durable local SSDs. To do this, AppOS comprises a Linux-compatible file I/O stack including virtual file system, page cache, block I/O layer, cloud storage driver. On top of the file I/O stack, syscall module supports registering pre- and post-handler for file I/O-related system calls in order to transparently work without modifying PostgreSQL codes.

I will focus on presenting key use cases and performance results of the AppOS extension after explaining the internals. Specifically, I will show the performance results of OLTP and some batch workloads using standard benchmarking tools like pgbench and sysbench. I will also present performance results and implications on multiple clouds including AWS, GCP, and Azure.