snowsam
join:2001-04-11
Signal Mountain, TN | 2 Clustered Switches for Equalogic 6100 + servers Since there was the Cheap switch thread going on ( »Recommend cheap 24 port managed gigabit switch ), I figured it was a good time to throw out what I am currently considering and see if anyone has some good advice.
We are implementing a Dell Equallogic 4100 and several host servers. This is the production environment for running a 75 person company.
I am looking for a switch to cluster. Dell has an approved list on page 9 of the PDF at »en.community.dell.com/techcenter···rix.aspx
On the 1Gigabit Ethernet Switches, for the Dell brand, the following are listed:
PowerConnect B-Series RX
Force10 C-Series
Force10 S25N
Force10 S50N
Force10 S55
Force10 S60
PowerConnect 6224
PowerConnect 6248
PowerConnect 7024
PowerConnect 7048
PowerConnect M6220
PowerConnect M6348
I have a quote for a cluster of two Force10 S25N switches and I have a quote on a cluster of PowerConnect 6224. There is a $2300 cost addition to go to the clustered Force10 S25N over the PowerConnect 6224.
The force 10 has a 5 year warranty. The PowerConnect has a lifetime.
Am I OK saving the money and going with the PowerConnect 6224 cluster?
Does anyone have any experience? |
|
JTY
join:2004-05-29
Ellensburg, WA | We use to use Dell 6224 switches in a stack with our EqualLogic arrays. Never had any issues with it. We've since switched to Force10 S50 switches, only because we are standardizing on them for Top of Rack as well. Either way works fine, in my experience. |
|
 zongPremium
join:2005-07-21
Scarborough, ON | reply to snowsam
Same here. A stack of 6224's feeding a PS4100, no problems whatsoever. |
|
 Badger3kWe Don't Need No Stinkin BadgersPremium
join:2001-09-27
Franklin, OH | reply to snowsam
3rd a stack of 6224s. I'm pretty sure the Force10 stuff is way overkill just for iSCSI traffic.
--
Team Discovery: Project Hope |
|
snowsam
join:2001-04-11
Signal Mountain, TN | Guys, thanks sooo much for the feedback. |
|
JoelC707Premium
join:2002-07-09
Stone Mountain, GA
kudos:5 | reply to snowsam
Hey look, I'm famous LOL
I've been doing a lot of research on the switches I'm considering in my thread and some discussions I've come across are on iSCSI implementations. Most everyone on other forums have recommended 6224's as well (if they are comparing Dell switches anyway).
That said, here's what I've come across for specs on good iSCSI switches. You want a switch that can handle flow control if nothing else. Jumbo frames are good and finding a switch that can handle both at the same time is ideal but flow control is preferred if you have to pick between the two. |
|
 fcislerPremium
join:2004-06-14
Riverhead, NY | reply to snowsam
6224/6248 are fine switches. Currently using two 6248 for an old PS4000 as our 10GB switch was out of 1G ports.
Flow control (on this switch, as of latest firmware) is a global on/off. Dell best practices not want it disabled on trunk ports. Ignore that.
To stack or not to stack?
Pro's of stacking:
-Single management interface
-Ability to create a port channel between both switches
Disadvantages:
-When you update you must bring the whole stack down
I have one setup of Force 10's that are not stacked and a dell 8024 that are stacked. If I had to do it again I would NOT stack. When I do an upgrade now I have to fail over to the backup switches first and then reboot the dells. With the force 10 I can upgrade one, verify it works, and then upgrade the other. Dell really needs to fix that - although I believe most other manufacturers (the Force 10 included) will have that same issue with a stack. |
|
 tubbynetreminds me of the danse russePremium,MVM
join:2008-01-16
Chandler, AZ
kudos:1 | said by fcisler: Dell really needs to fix that - although I believe most other manufacturers (the Force 10 included) will have that same issue with a stack.
as of a few software revisions ago (somewhere in the 12.2(50-something) base) -- cisco implemented rolling upgrades on stacked switch members for the c3k.
however -- in most cases -- the best solution is to run a switch that supports some form of mc-lag -- wherein the control-planes between the members remain separate yet provide a "pseudo-unification" at the dual-homed device (dhd) end so that things like lacp are supported.
q.
--
"...if I in my north room dance naked, grotesquely before my mirror waving my shirt round my head and singing softly to myself..." |
|
