Poor perfomance with 3660 via FC

[Stefano Colombo]

We're installing a storeonce 3660 FC and noticed quite a poor performance during network-free backup.

The setup is as following:

Primay storage alletra 5000 16Gb FC

Backup Gateway server 8Gb FC

Storeonce 3660 16Gb FC

Veeam 12.2.

The storeonce has been configured as repository with FC activated and the gateway server can reach it correctly, SAN has been zoned.

The JOB is configured with storage integration and it creates the Alletra snapshot mounting it to the gateway server.

Given such configuration the backup job is running at about 200MB/s where the same job using a diskbased target via FC as well is running at 1GB/s, which make sens of the 8Gb gateway server's HBA

The SAN Fabric is made of 2 switches, one is an 8Gb Blade switch and the second is a SN3000 16Gb switch.

Th Gateway server is connected to the Blade Switch while the Alletra and the Storeonce are connected to the external 16Gb switch

[Cali]

Hi,

in general i see higher performance if the StoreOnce is connected by LAN as by FC.
If you have FC, try to configure more connections (20 is the default).
You can see the Number in the Windows Device Manager as "Unknown Devices".

See "HPE StoreOnce 3660, 5260 and 5660 Systems User Guide for StoreOnce" search for "StoreOnce Catalyst Fibre Channel number of devices per login".

Try with a higher Number, until you get no more Performance.

Also, remember that StoreOnce gets higher results with multiple concurrent backup sessions.

Cali

[Sunitha_Mod]

Hello @Stefano Colombo,

Let us know if you were able to resolve the issue.

If you are satisfied with the answers then kindly click the "Accept As Solution" button for the most helpful response so that it is beneficial to all community members.

[Gans_Tapes]

@Stefano Colombo  

Performance cases have a lot of factors to be reviewed. I'd have more questions for you than answers.   

1. Was this a working setup? If yes, when was it working ok and what was the performance?

2. If this is a new setup, then what is the performance numbers we are monitoring and where are we observice the same? (From the  backup application OR from the StoreOnce telemetry dashboards etc or the gateway servers or the traffic at the switches) 
3. As @Cali  rightly pointed, if running FC, then some basic checks include ensuring the FC switches are at the supported firmware versions, the StoreOnce to be at the latest and ensuring there are no cable/sfp issues between the source and target.

I suggest you get a case logged for the StoreOnce system to investigate the same as we need to understand the environment before we could provide appropriate recommendations. 

[Stefano Colombo]

EDIT:

In that specific case the problem has found to be the ISL between the two FC switches, as strange as it seems.

We moved the storeonce on the same switch connected to the veeam physical proxy host and then the performance went immediately up to 1GB/s

Hope this may help

[Mohammed_I]

Hi Stefano Colombo,
Thank you for reaching HPE community.
For Performance cases have a lot of factors to be reviewed. 
In this case we need logs and more details to find the bottle neck.
My recommendation is to log a Support case with Store once Team.
StoreOnce Engineer will check the logs and then update you solution .

Thank you. HPE Community team.