Storage Boards Cleanup
To make it easier to find information about HPE Storage products and solutions, we are doing spring cleaning. This includes consolidation of some older boards, and a simpler structure that more accurately reflects how people use HPE Storage.
Disk Arrays
Showing results for 
Search instead for 
Did you mean: 

EVA 6000 i/o problem

Occasional Advisor

EVA 6000 i/o problem

We have i/o problem on EVA 6000.

-2TB oracle DB on rx6420.
-File systems on EVA 6000. 2T, 1.5T, 2TB. FATA(7.2K, 500GB)
-Separate FS is used for Oracle Flash Recovery.

The copy of 1 GB data on EVA takes more than (120 sec).

Similar server that works on FC (10K, 300GB) is five times faster.

I am in critical situation as full DB backup takes more than 20 hours.

Here is the output of iostat command.

device bps sps msps
c10t0d3 3001 100.0 1.0
c11t0d3 0 0.0 1.0
c6t0d3 0 0.0 1.0
c7t0d3 0 0.0 1.0
c10t0d4 0 0.0 1.0
c11t0d4 0 0.0 1.0
c6t0d4 0 0.0 1.0
c7t0d4 0 0.0 1.0

Thank you.
Ivan Ferreira
Honored Contributor

Re: EVA 6000 i/o problem

Your performance metrics are not enough to identify if a problem occurs, this is because:

Writing data to a file system is not a good performance metric because writes are cached, you should write directly to raw devices if you want to do performance test (only if you have a lun without data, a test lun).

The server load can influence in your performance metrics, you must run the test without any other activity on the server.

The iostat you provided shows little activity, and we don't know the sample number and interval.

We don't know your Tape device speed and the type of backup (full, incremental, file sytem based, RMAN based).

We don't know the real ammount of space used.

We don't know how many disks you have in the Storages or RAID level.

You are comparing 7.2k vs 10k drive speeds.

If you backup 5.5 TB in 20 Hours, you are getting about 80 MB/s, this is normal for an LTO3 DRIVE.

If you are getting full speed of your tape drives, and your backup window is too long, you must redesing your backup strategy.
Por que hacerlo dificil si es posible hacerlo facil? - Why do it the hard way, when you can do it the easy way?
Occasional Advisor

Re: EVA 6000 i/o problem

Thank you Ivan.

Let me explain my problem in details.

You are right that iostat is not measure. That's why i wrote about 1GB file. I copied it many times. The result was the same - about 120 sec, even ORACLE was down.

At oistat, only activity was data transfer to the tape library (SUN L180, LTO2, 2 Drives). The speed from other stations is 12MG/s (LAN limit). For EVA 2-3 GB/s.

Oracle backup is taken by RMAN. RMAN disk COMPRESSED backup. I am using 8 CHANNELS and PARALLELISM 4. It is also determined by experiments.
After RMAN finishes, backupsets are sent to the tape library.

DB size is 2T not 5.5T.
2TB used by Oracle Database. 2TB recovery area where RMAN backupsets are stored. 1.5T free space. So, I have 2TB for 20 hours. This disk backup not tape. For my tape library it would be the great speed :).

Unfortunately we are not using raw devices for ORACLE.

About Disks and RAID. There are two units with 14 FATA (500 GB) on each. Also, I was told that RADI51 or RAID01 combination is configured. But I am not sure.

Occasional Advisor

Re: EVA 6000 i/o problem

Any suggestion ?