2.2K Posts

February 25th, 2009 10:00

Have you tried instead spreading the LUNs across multiple consistency groups and then using a Group Set to create write-order consistent bookmarks across all the consistency groups?

Also is your change rate so high that you cannot even perform the first full synchronization?

We are a SQL shop and the biggest SQL databases I replicate is a 4TB OLTP database with high transactional rate. We solved the throughput issue by using three consistency groups and spreading the LUNs that comprise the database across all three groups. Then we created a Group Set that creates write-order consistent bookmarks across all three groups.

257 Posts

March 2nd, 2009 00:00

Hi Goran

Aran is absolutely right, a group set is a bookmark which spans multiple consistency groups, so you can have write order fidelity (crash consistency) across those CGs in that groupset bookmark.

They are very useful in harnessing the power of available RPAs for high change rate databases.

Now, on to your info about the 100mbits link and only doing 50-60mbits...
A Gen-3 RPA can handle up to ~60 Mbytes/sec, so saturating a 100mbit link should be no issue here.

Just remember, during a syncronization, the RPA will read up from both source and target LUNs as fast as they can support and only ship what data is required across the link. So, depending on how busy your LUNs are already, the LUNs might only be able to service X amount of mbits to the RPAs.

But again, this is speculation... I suggest you open a call with customer service and they can take a look at the situation and ensure you are getting the best from your RecoverPoint solution.

Cheers
James
No Events found!

Top