Dr Group Log Size

[Gpapa]

Hi All,

Have a quick question in DR Group Log Sizes.

I have a EVA4400 in a second location that im am using Continous Access to replicat data to.

The EVA has 8 x 400GB hard drive
Single Drive Protection
6 DR Groups and a
690 GB of vDisks (all raid 5, including local, replicated and snapclones)
2233.66 GB total space
169.49 GB Available

now is all the space gone ?

I see that each DR group has 102,400 MB of Log space, I have been told me that the space is reserve at each end in Raid 1.
Is this true?
am i losing more then half my total space (1200GB)to these Logs?

[Uwe Zessin]

Welcome to the club...

Starting with XCS 09003000 the default WHL size indeed seems to be 100 GigaBytes per Data Replication Group with a fixed allocation.

[Gpapa]

This is crazy

I have migrated 790 GB out of 1350GB from my EVA3000 (6 out of 8 DR Groups) and im out of space on my new EVA.

the new EVA has 400GB more space then my EVA3000.


I am wondering if maybe i should delete all but 1 DR Group add the members to that group

that would save me 1400GB

[Uwe Zessin]

Remember that all virtual disks belonging to one Data Replication Group must be maintained by the same controller.
I can't test it right new - can you specify a smaller WHL size?

[Gpapa]

Log Size cannot be changed after the DR group is created

When creating the DR Group it can be set

"This is the maximum size for the DR log disk. The value is the number of blocks in 512 bytes. If left blank or set to 0, the controller software calculates the best log size for the space available. Otherwise, you can specify a size, which must be between 136 MB and 2 TB (278,528 through 4,294,967,295) entered in blocks."

How do you calculate how big to make the Log Size though?

[Uwe Zessin · Accepted solution]

> How do you calculate how big to make the Log Size though?

You have to guess how long the interlink between the EVAs will be down at maximum and how much data will be written during that time. Watch out: the WHL is a true 'journal' - if a single block is written 1000 times, all writes are stored in that log.

[Gpapa]

Some of my DR Groups only contain 30 GB of members, so 100 Gig seams like over kill.

What happen if the the log file fills up during the a link outage ? is the log file used when the Replication is suspended?

[Uwe Zessin]

If there is no more space for logging, the Data Replication Group will automatically do a 'full-copy' after the interlink starts working again (I have seen this really happen on two EVA3000). Of course, the data is not consistent on the destination vdisk as long as the full-copy runs...
If you manually suspend a Data Replication Group, yes, the group should start logging.

[Gpapa]

Thank You Uwe

You have given me alot to think about

unfortunately it appears the is no clear answer

I may pause Replication for a few hours to see how big the log file gets before i decide on how big they should be.

[Uwe Zessin]

You're welcome.

Unfortunately, every system is different and even the load can change from time to time. If there were 'on correct answer', the engineers could have set it for you already.