Sysadmin

8992 readers

5 users here now

A community dedicated to the profession of IT Systems Administration

No generic Lemmy issue posts please! Posts about Lemmy belong in one of these communities:
[email protected]
[email protected]
[email protected]
[email protected]

founded 2 years ago

MODERATORS

[email protected]

What to do when a MySQL/MariaDB database gets too large for a single host? (self.sysadmin)

submitted 2 days ago by otto to c/[email protected]

7 comments fedilink hide all child comments

What are your strategies when a MySQL/MariaDB database server grows to have too much traffic for a single host to handle, i.e. scaling CPU/RAM is not an option anymore? Do you deploy ProxySQL to start splitting the traffic according to some rule to two different hosts? What would the rule be, and how would you split the data? Has anyone migrated to TiDB? In that case, what was the strategy to detect if the SQL your app uses is fully compatible with TiDB?

you are viewing a single comment's thread
view the rest of the comments

[–] [email protected] 1 points 2 days ago* (last edited 2 days ago)

I apologize for not providing a link. (https://en.m.wikipedia.org/wiki/Altix) I am not quickly finding specs.

These were SGI Altix systems before HP bought SGI. Tightly integrated clusters such that they operated as a single NUMA space. They are/were often used to host databases with massive shared memory.

The smaller systems had lower numbers, and older had numerically lower numbers. A UV 1000 wad two models previous to the UV 3000. The UV 100 was same generation as 1000, but smaller.

If I recall correctly, the UV 100 had 3 TiB RAM. These are very old now, and only an example. The UV 3000 had way more RAM and CPUs.

A modern single non-UV server maxed out can hit over 1 TiB (I have not spected one in a while). Expect the single server to cost over $20K, anything less means one is in the wrong section of the store.

Edit: clarifying a point

Edit2: Just checked and a single server can hit 3 TiB RAM with 128 Cores for around $53K. Put that $53K in comparison with employee time for any other solution.