Thu 29 May 2008
Well as meetings progress and executives sit impatiently, our D.R. plan is slowly coming together. Having to test and choose products is always a great learning experience, though doing this with deadlines takes some of the fun away.
We’ve already sat through the presentations from Lefthand and Equallogic since we’ve narrowed our choices for our production SAN down to these two venders. Just from meeting the reps from both of these companies, I’d have to say that the Lefthand guys seem more down to earth and know what the hell they are talking about. The Equallogic reps we’ve had out are just salesmen who have had minor hardware training and only answer questions with generic answers. The Lefthand guys do this for a living and for fun it seems, the Equallogic guys are just salesmen, period.
One of the biggest downsides for us with Equallogic was that when it comes down to upgrading the firmware on the modules, it will take down network connectivity for 15 seconds while it switchs over to the secondary controller inside the array. I don’t know about other companies and how they would handle their SAN being down for 15 seconds a couple times a year but I’ll pass. With Lefthand, because of their network RAID design, no downtime is required. I’d have to say we’re leaning very heavily towards Lefthand for the SAN purchase.
We’ve settled on VMware 3.5 for our virtualization needs. Virtual Iron will hopefully make it into our plans some day but they just aren’t mature enough for what we need. I’d like to actually use the licenses we’ve purchased for Virtual Iron to create a testing lab for our dept since we just don’t have space for phsyical machines.
And I finally found the time to install and configure Cacti to replace Servers Alive. The newest release from CactiEZ is really the best and easiest solution i’ve found for the price paid, FREE! The features that cacti has and the plugins available really do compare to most enterprise solutions that cost a bucket full of cash. I have almost all of our servers setup and am going through networking equipment to add as needed. In the past if a developer said they needed a new server because a current server just couldn’t handle the load, we would just buy it. Now with performance monitoring and tracking I can ask them to prove the need. In the past month i’ve had to pull statistics on a few servers for this exact reason.
4 Responses to “More meetings….”
Leave a Reply
You must be logged in to post a comment.
May 29th, 2008 at 6:12 pm
Hi, Marc Farley with Dell EqualLogic. Sorry to hear about your lackluster experience with our sales folks. I’ve contacted who I think they are and hopefully they will be contacting you again soon.
Obviously, the architectures of EqualLogic and Lefthand’s products are different. When a Dell EqualLogic system is getting its firmware updated it goes offline for 15 seconds - just like nearly all enterprise midrange systems do. (there is some period of time when the systems stop - and stop accepting I/Os).
FWIW, this is much smaller than the amount of time it takes failover to occur if you are using multi-pathing. SCSI time outs are implemented in host systems for these types of things to resolve without problems. Many people experience this type of delay every day without a hiccup, because its the way server I/O is designed to work.
The thing the Lefthand people probably didn’t explain was the fact that they don’t have a way to turn off write caching when updating their system. Because there are multiple single points of failure in a Lefthand system, any one of those things could take a hit while an upgrade is going on with their other system (and does it take 5 minutes - 10 minutes? - I really don’t know) and you run the risk of losing data that was never flushed from cache. I’m sort of surprised they took you down this road. Of course, you can have three Lefthand systems, but the cost starts looking a lot less attractive.
Also, if you have SQL Server and Exchange, you really need to look at the snapshot recovery capability we have compared to Lefthand. We have crash consistent snaps for both. Lefthand doesn’t. In general, the whole snapshot restore scenario is much better with EqualLogic. If you restore a snapshot with Lefthand, you lose your other snaps. Not so with EqualLogic.
May 30th, 2008 at 8:29 am
Luke..
Attracting the big dogg to dinner huh ?
Ah the wonders of the internet.
It’s funny how the bigger the dogg, the bigger the pile sometimes left on your lawn!
May 30th, 2008 at 4:50 pm
And Zirkdaddy is from Lefthand?
June 18th, 2008 at 12:30 pm
I have been demo’ing a pair of PS arrays. I was ready to remove Equallogics from our short list of storage vendors because of skepticism about the firmware reboot requirement. Luckily for Dell our project got an extension and I decided to give them a shot. I have been testing a boot-from-san Windows 20003 server using two Qlogic QL4050C adapters with PowerConnect 5448 switches. After getting the various switch and Server hardware / software settings tuned I have been pleasantly surprised with the testing results.
Tests involved copying over the network a 10GB Microsoft IMG file to the boot-from-san server then yanking storages controllers, rebooting the array and even upgrading the firmware to the newest release, all had no interruption to the host or the file copy process. I did not even see an event in the server’s eventlog.
A couple of Notes:
I timed failover storage controller after abruptly removing the active storage controller at about an average of 50 seconds I performed this test 5 times all with no host outage.
A controlled array reboot took 15 seconds again with no host outage.
I only have one server on the Array
I am using one member in the PS peer group.