convention. When sharded, a keyspace has N shards with non-overlapping data. Resharding Vitess supports dynamic resharding, in which the number of shards is changed on a live cluster. This can be either splitting one or more shards into smaller pieces, or merging neighboring shards into bigger pieces. During dynamic resharding, the data in the source shards is copied into the destination shards, allowed to catch bit set will go to -80. All others will go to 80-. Applying the above change should result in the creation of six more vttablet pods, and the following new jobs: NAME DESIRED SUCCESSFUL AGE jobs/customer

SetOwnerColumns function. If present, the vschema builder will call this function at the time of creation and supply the column names of the owner table. This will allow the Create to issue the select for it’s converting from/to. 61 Supported character sets The list of supported character sets is dynamic and may grow. You will find it under CharacterSetEncoding in https://github. com/vitessio/vitess this would make it very hard for monitoring tools because they are not very good at dealing with dynamic variable names. 87 HTTP endpoints A TabletServer exports a variety of http endpoints. In general

team at YouTube faced. This section briefly summarizes the sequence of events that led to Vitess’ creation: 1. YouTube’s MySQL database reached a point when peak traffic would soon exceed the database’s management GUI – Designed to work in multiple data centers / regions • Sharding – Virtually seamless dynamic re-sharding – Vertical and Horizontal sharding support – Multiple sharding schemes, with the ability (implemented in different ways by Oracle, SQL Server and PostgreSQL). • Very high throughput for ID creation, using a configurable in-memory block allocation. • Transparent use, similar to MySQL auto-increment:

SetOwnerColumns function. If present, the vschema builder will call this function at the time of creation and supply the column names of the owner table. This will allow the Create to issue the select for this would make it very hard for monitoring tools because they are not very good at dealing with dynamic variable names. HTTP endpoints A TabletServer exports a variety of http endpoints. In general, team at YouTube faced. This section briefly summarizes the sequence of events that led to Vitess’ creation: 1. YouTube’s MySQL database reached a point when peak traffic would soon exceed the database’s

SetOwnerColumns function. If present, the vschema builder will call this function at the time of creation and supply the column names of the owner table. This will allow the Create to issue the select for this would make it very hard for monitoring tools because they are not very good at dealing with dynamic variable names. 62 HTTP endpoints A TabletServer exports a variety of http endpoints. In general team at YouTube faced. This section briefly summarizes the sequence of events that led to Vitess’ creation: 1. YouTube’s MySQL database reached a point when peak traffic would soon exceed the database’s

SetOwnerColumns function. If present, the vschema builder will call this function at the time of creation and supply the column names of the owner table. This will allow the Create to issue the select for this would make it very hard for monitoring tools because they are not very good at dealing with dynamic variable names. HTTP endpoints A TabletServer exports a variety of http endpoints. In general, team at YouTube faced. This section briefly summarizes the sequence of events that led to Vitess’ creation: 1. YouTube’s MySQL database reached a point when peak traffic would soon exceed the database’s

team at YouTube faced. This section briefly summarizes the sequence of events that led to Vitess’ creation: 1. YouTube’s MySQL database reached a point when peak traffic would soon exceed the database’s management GUI – Designed to work in multiple data centers / regions • Sharding – Virtually seamless dynamic re-sharding – Vertical and Horizontal sharding support – Multiple sharding schemes, with the ability (implemented in different ways by Oracle, SQL Server and PostgreSQL). • Very high throughput for ID creation, using a configurable in-memory block allocation. • Transparent use, similar to MySQL auto-increment:

team at YouTube faced. This section briefly summarizes the sequence of events that led to Vitess’ creation: 1. YouTube’s MySQL database reached a point when peak traffic would soon exceed the database’s management GUI – Designed to work in multiple data centers / regions • Sharding – Virtually seamless dynamic re-sharding – Vertical and Horizontal sharding support – Multiple sharding schemes, with the ability (implemented in different ways by Oracle, SQL Server and PostgreSQL). • Very high throughput for ID creation, using a configurable in-memory block allocation. • Transparent use, similar to MySQL auto-increment:

s/vitess/fuzz_keyspace_creation .go Yes 2 FuzzShardCreation https://github.com/cncf/cncf-fuzz ing/blob/83bad32323d4a351571 7c5f144faf38b2c7d20cb/project s/vitess/fuzz_shard_creation.go Yes 3 FuzzTabletCreation

defers • generics • GC • map, slices • performace • java < go < c/c++ Vitess Features • dynamic resharding • auto-failover • row-cache • limit inefficiency SQL • more client connection • Structure Vitess Resharding • read scaling up ? —> replica • write scale up ? —> split • dynamic shard • scale • less than 5 seconds read-only downtime Vitess Resharding • shard0, shard1