Darin Strait on 14 May 2008 07:06:22 -0700

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: [PLUG] real-time data replication over highspeed WAN link


I''ve spent a lot of time working with SQL Server running on big EMC hardware, and I've got some understanding of how their SAN-based replication tech (SRDF) works with SQL Server.

As long as whatever SAN you settle on looks just like any other disk to windows, SQL should be happy enough. Either iSCSI or a "real" SAN" shouldn't matter, at least from a getting-it-to-work perspective. A well-peforming system is a different question. "Network drives", the kind that you get off of NAS units and are available through UNC-style paths, are usually a bad way to go.

The final arbiter of a high-performance SQL Server system is I/O latency, particular for SQL Server's tempdb files and the transaction log files for the user databases. SQL Servers are particularly sensitve to disk latency. This sensitivity can kill performance inside of a VMWare VM, for example. If your database is low-stress, it might not matter much, or even noticeably, but when  organizations place their databases files on SANs, performance often drops. Many people are surprised by this because marketting teaches us that SANs are faster.

One thing, from your note, it's not clear what purpose you intend to use the 'other' copy for. SQL Server doesn't support two instances running off of a SAN-based replication system. There is no Oracle RAC-like feature and "clustering" is often not what people are looking for, either. People who aren't "SQL Server guys" don't always realize this. I don't know what your background is, so I'm just sayin'.

The other copy would be only for failover/disaster recovery purposes. (Of course, you'd want to test the way your system works, as implemented, by knocking down one server and bringing up the DR server.)

At the SQL Server level, your choices for data "replication" (by which I mean a copy of the data) are "log shipping", "database mirroring" and SQL Server's internal replication features. All of these have trade offs, which may or may not be important to you, depending on exactly what you need.

-- d

On Tue, May 13, 2008 at 9:40 PM, W. Chris Shank <shankwc@acetechgroup.com> wrote:
I'm wondering if anyone has used any data replication tools like DRDB for high availability fail-over through a WAN link. I am considering the Dell EqualLogic iSCSI SAN with it's peer replication technology. But so far it seems like a pretty hefty price-tag for something I could accomplish with Linux, DRDB, and iSCSI tools. Does anyone have a failover site with real-time replication? If so, what are you using? My environment is mostly Windows Servers and MS SQL Databases - so I was only considering Linux because it seems like I could replace the EqualLogics with a Linux SAN. But I'll give consideration to other hardware or software solutions as well.

My biggest concern is if I can keep the databases in sync in real-time (if not real-time - how close), what bandwidth will I need to have been datacenters, and will the replication give me a performance hit at my primary location. Your suggestions are appreciated.

W. Chris Shank 
ACE Technology Group 

Philadelphia Linux Users Group         --        http://www.phillylinux.org
Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce
General Discussion  --   http://lists.phillylinux.org/mailman/listinfo/plug

Philadelphia Linux Users Group         --        http://www.phillylinux.org
Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce
General Discussion  --   http://lists.phillylinux.org/mailman/listinfo/plug