Tech, ExplainedData platforms and AI
Product

Open-source database support

Who to call when PostgreSQL slows down.

BEFORE ITAFTER ITOpen-sourcedatabase supportPerconaCOVERS24x7 helpQuery tuningUpgradesMonitoring toolsRelational databasereportsslow queriesDocument databasereportserrorsThe database is back at normalspeed, with the cause found andthe fix written upCONCEPTS IT USESOpen source business modelBEFORE ITAFTER ITOpen-sourcedatabase supportPerconaCOVERS24x7 helpQuery tuningUpgradesMonitoringtoolsRelationaldatabasereportsslow queriesDocument databasereportserrorsThe database isback at normalspeed, with thecause found andthe fix written upCONCEPTS IT USESOpen source business modelBEFORE ITRelational database↓ reports slow queriesDocument database↓ reports errorsOpen-source database supportPercona24x7 helpQuery tuningUpgradesMonitoring toolsAFTER IT↓The database is back at normal speed, withthe cause found and the fix written upCONCEPTS IT USESOpen source business model
Open-source database support receives slow-query reports and errors from relational and document databases a company runs itself, and ends with the database back at normal speed.
opens a page inside this one a page in the same scene a page in another scene a concept page a label only, with no page

Open-source database support is a paid service in which outside experts help a company run free, open-source databases, such as PostgreSQL, MySQL and MongoDB, on its own servers, and fix them when they break or slow down. The seller usually adds tools for monitoring, backup and security, and advice on upgrades.

Open-source software can be downloaded and run free. But with free software, there is no company that must answer the phone when it fails at 2 a.m. That is the gap this service fills. It is one of the three ways companies earn from free software, as the open source business model explains.

A support contract usually promises a response time, for example within one hour for a database that has stopped. The seller's engineers know the database's internals: how it plans a query, how it stores data on disk, and how it copies data to a standby server. Many sellers also offer monitoring tools, their own tested builds of each database, and consulting for upgrades and moves from one database to another.

Here is a typical case. A concert ticketing platform in Chandigarh runs PostgreSQL, a relational database, on its own servers. Tickets for a popular singer's concert go on sale at noon, and three lakh people open the site within minutes. The query that checks free seats, which normally takes 40 milliseconds, starts taking 6 seconds, and pages time out. The company calls its support provider. Within half an hour, the support engineer finds two causes. The seats table has no index on the column that the query searches, so the database reads the whole table every time. An index is a sorted list of one column's values that lets the database go straight to the right rows. And the app servers have opened 500 connections, the most the database allows. The engineer adds the index and a connection pooler, a program that lets many app requests share a few connections. The query is back to 50 milliseconds. A week later, the provider recommends an upgrade, because the PostgreSQL version in use will soon stop receiving security fixes.

Support covers other databases too, such as a document database that runs out of memory.

Why not simply rent database as a service (DBaaS)? Some companies want full control of their servers. Some must keep data in their own data centre. Some find that running the database themselves costs less at their size. And some want to avoid vendor lock-in to one cloud provider's database. These companies buy support instead, and keep running the database themselves.

Tech, Explained is a part of ingrid.fyi. Every page explains one piece of technology in plain words.