Message Boards

Why we need Liferay clustering?

Maulin Patel, modified 3 Years ago.

Why we need Liferay clustering?

New Member Posts: 9 Join Date: 6/16/20 Recent Posts
1. Why we need Liferay clustering. Instead, can I use one DB for two different servers?
2. I am aware of Lucene cache will be the problem but what if I disable Lucene cache?
3. Someone in this thread tries to disable the cache but got some error. What was the issue and who can I resolve it?
https://liferay.dev/forums/-/message_boards/message/16212310
thumbnail
Olaf Kock, modified 3 Years ago.

RE: Why we need Liferay clustering?

Liferay Legend Posts: 6403 Join Date: 9/23/08 Recent Posts
Maulin Patel:

1. Why we need Liferay clustering. Instead, can I use one DB for two different servers?
2. I am aware of Lucene cache will be the problem but what if I disable Lucene cache?
3. Someone in this thread tries to disable the cache but got some error. What was the issue and who can I resolve it?
https://liferay.dev/forums/-/message_boards/message/16212310

1. No. Because you'll run into issues.
2. There's no more lucene since version 7.0, and you shouldn't run any older version by now
3. Who cares? As soon as that issue is solved, you'll run into the next one.
If you have two or more servers serving your load, just configure a proper cluster. That's significantly less effort than going any other route.
thumbnail
David H Nebinger, modified 3 Years ago.

RE: Why we need Liferay clustering?

Liferay Legend Posts: 14919 Join Date: 9/2/06 Recent Posts
Maulin Patel:

1. Why we need Liferay clustering. Instead, can I use one DB for two different servers?
2. I am aware of Lucene cache will be the problem but what if I disable Lucene cache?
3. Someone in this thread tries to disable the cache but got some error. What was the issue and who can I resolve it?
https://liferay.dev/forums/-/message_boards/message/16212310


The "lucene cache" only affects the index. There's also hibernate-level caching provided by EhCache to hold retrieved entities.


The Liferay clustering sends out a cluster-wide notification during saves so any nodes holding the now-stale entities will eject them from the caches and re-retrieve as necessary.

And before you say "Oh, I'll just disable those too", that means you'll be turning every data request into a DB request which which dramatically impact overall performance. Liferay is tuned to leverage caching to reduce overhead; eliminating cache will eliminate all of that tuning, performance will suffer as a result.