Storage Essentials Practitioners Forum
cancel

"Highly Available" Storage Essentials?

Highlighted
CA1471374
New Member.

"Highly Available" Storage Essentials?

Since one cannot cluster SE 5.1.3, I am looking for anyone who may have advice/opinions about making an SE environment "highly available." Consider two EVA8x00 cells FC connected via CWDM. The goal would be to somehow synchronize the "live" SIM/SE instance (two separate servers on each side) with "hot" standby servers at the remote location. In the event on a disaster, the SE/SIM instance would be available with fairly accurate data. I have considered using block-level replication over IP in addition to the risky proposition of putting SE/SIM on the SAN itself and CAing those LUNs. A colleague noted that it seemed strange to house the data on the very element that you may need to "control" in the event of a disaster and therefore recommends using only local storage.

Does anyone do this? Is there a need to do so? I'm wondering how important my SE environment is going to be in the event of a disaster; after all, I have CV to access my EVAs.

Any opinions on this?
4 REPLIES
James Muell
Regular Contributor.

Re: "Highly Available" Storage Essentials?

Great question.
We have run Appiq (now SE) for 2 + yrs now. We originally askes about an HA solution but it was not on the roadmap (it may be now the HP owns it) My thought was since it is backended by an Oracle instance, to setup an Oracle cluster as the 1st line HA solution. This could conceivably be a GEO cluster, though there will inevitably be latency to deal with. There are other Oracle methods for pushing and pulling data from one instance to another that might be useful in the DR sense. If a primary site is lost you would have a copy of neartime data.
We also have a 2nd site running SE and have thought about using a.) SE's Global Reporter to pull both sites data, but my understanding of GR is it only pulls an portion of the data.
Stephen Kebbell
Super Contributor.

Re: "Highly Available" Storage Essentials?

Hi,

I have heard that sometime in the future it is planned to separate the DB and Application parts of SE. So you could, for example, run the Oracle DB on a Cluster, and the App-Server on another Windows Server (maybe even a Virtual Server on ESX). But I have no definite info. Try talking nicely to your HP rep, and see if they could show you the roadmap for SE.

We have another vendor's version of SE, and it's a standalone server with local disks. I consider it not to be critical in case of a disaster. We still have the native array management tools for that.

Regards,
Stephen
Amiyanji
New Member.

Re: "Highly Available" Storage Essentials?

I am planning to install SE V6.0. Does anyone know if this version can be configured in High Availibity Mode, if yes how?.
James Muell
Regular Contributor.

Re: "Highly Available" Storage Essentials?

I have been running 6.0.1 and sp.4 beta (vmware friendly) for a few months. I have been after them for years - before Hp bought it from APPIQ. They don't intend to at this time. My config is SE w/oracle on host-A and SIM5.2.x on host-B. There is a "Connector" component to map devices etc btwn Oracle and SIM/SQL db. Using SIM has advantages and dis-advtges. If configured correctly SIM will do all the discovering for you. There is no getting away from the DB.
-- Do you need HA? Since it is not in the FC path btwn host and storage sys, its not really critical as a mgmt tool. It can be restored by tape bkup. The app also lets your capture all host info for recovry. Criticallity comes in play when you, or other apps are dependent on the data - nightly change data - for some other purpose. Running 5.1.3 in production has been difficult and we're always having to reset the temp space. They say 6.x is better. At HP Softwr Univ - a session with their dev team - all customers leaned heavily on driving the performance of the application.