No, the array would only fail over if there was a hardware failure on the control module.
Your iSCSI network connections from the Array to the switches are not setup properly for High Availability. The proper network setup for high availability is to have a minimum of 1 port from each array controller plugged into each switch; In your example, you would need CM0 eth0, 1, & 2, plugged in to SWA and CM0 eth3 on SWB. The do the opposite for CM1, eth 0, 1, 2 is plugged into SWB, where CM1 eth3 is in connected to SWA. Note I say at least one eth interface as the “barest” minimum (i.e., if you did have a switch failure, your network performance could be degraded with just one active interface on the array (depending on time of day, network load, etc.). I also suggest that you fully test your fail over to ensure that all hosts and both switches can ping all interfaces by testing a hardware failover on the array and pinging/tracert all active interfaces on each controller (test with CM0 active, the retest with CM1 active).
Here is a link to building a High Availability iSCSI Network:
http://attachments.wetpaintserv.us/CjSA1VnfNTHzy7O03UbN0g485148
With the configuration you have called out, if you suffer a switch failure you can maintain access to the EQL group albeit with degraded performance. While this configuration does allow you to suffer a single switch failure while maintaining access with degraded performance...
A query I would have is that would it not be better to have all connections for each controller plugged into a dedicated switch, then if it fails, let the controller recognise that ports have failed using Link Status detection and then fail over to the 2nd controller which will have 4 active connections, therefore maintaining performance....?
Future product strategy can't be discussed in an open forum. You're best option, is to contact your sales folks, get an NDA (Non Disclosure Agreement) in place and they will be happy to discuss some of the features on the product roadmap.
Yeah, it would be good to have the option to enable link detection failover OR controller hardware failover. I'm kind of getting tired of having to explain to folks internally what conditions would invoke controller failover, and we've been bitten by misconfigured switch ports and bad cables where the passive controller is connected. Having a controller fail back to where there's a good link sounds like a really good idea to me.
Agreed on the link failure detection and would love to see this feature! Using two switches seems like a huge waste of bandwidth. Spanning tree will block the links on one of the two switches until a failure happens.
The new PS41xx/61xxx have this. If a switch or port fails that's connected to the active CM, it will check the status of the port on the passive CM and use that if available.
Is this true? I agree with the OP that network link detection-based failover would be an important feature enhancement to the EqualLogic series, but I do not see any cabling guidance for the PS4100/6100 series that differs from the PS4000/6000 series. The 10GbE has its own considerations because of the active/passive nature of the 10GbE NICs, but it looks like the cabling diagrams for a PS4100 still shows a "mesh" approach as best practice.
Cabling The Dell EqualLogic PS4000/PS4100/PS6010/PS6510 Storage Arrays 1. Connect a network cable from the network switch 0 to Ethernet 0 on the control module 1. 2. Connect a network cable from the network switch 0 to Ethernet 0 on the control module 0. 3. Connect a network cable from the network switch 1 to Ethernet 1 on the control module 1. 4. Connect a network cable from the network switch 1 to Ethernet 1 on the control module 0. 5. Repeat steps 1 to 4 to connect the additional Dell EqualLogic PS4000/PS4100/PS6010/PS6510 storage array(s) to the iSCSI switches.
Cabling The Dell EqualLogic PS6000/PS6100/PS6500 Storage Arrays 1. Connect a network cable from the network switch 0 to Ethernet 0 on the control module 1. 2. Connect a network cable from the network switch 0 to Ethernet 1 on the control module 1. 3. Connect a network cable from the network switch 1 to Ethernet 2 on the control module 1. 4. Connect a network cable from the network switch 1 to Ethernet 3 on the control module 1. 5. Connect a network cable from the network switch 1 to Ethernet 0 on the control module 0. 6. Connect a network cable from the network switch 1 to Ethernet 1 on the control module 0. 7. Connect a network cable from the network switch 0 to Ethernet 2 on the control module 0. 8. Connect a network cable from the network switch 0 to Ethernet 3 on the control module 0. 9. Repeat steps 1 to 8 to connect the additional Dell EqualLogic PS6000/PS6100/PS6500 storage array(s) to the iSCSI switches.
If this feature is available in these new models, can someone point me in the right direction for the documentation?
Joe S586
7 Technologist
•
729 Posts
0
May 6th, 2011 07:00
No, the array would only fail over if there was a hardware failure on the control module.
Your iSCSI network connections from the Array to the switches are not setup properly for High Availability. The proper network setup for high availability is to have a minimum of 1 port from each array controller plugged into each switch; In your example, you would need CM0 eth0, 1, & 2, plugged in to SWA and CM0 eth3 on SWB. The do the opposite for CM1, eth 0, 1, 2 is plugged into SWB, where CM1 eth3 is in connected to SWA. Note I say at least one eth interface as the “barest” minimum (i.e., if you did have a switch failure, your network performance could be degraded with just one active interface on the array (depending on time of day, network load, etc.). I also suggest that you fully test your fail over to ensure that all hosts and both switches can ping all interfaces by testing a hardware failover on the array and pinging/tracert all active interfaces on each controller (test with CM0 active, the retest with CM1 active).
Here is a link to building a High Availability iSCSI Network:
http://attachments.wetpaintserv.us/CjSA1VnfNTHzy7O03UbN0g485148
Here is a link to our configuration guide:
http://www.delltechcenter.com/page/EqualLogic+Configuration+Guide
Regards,
Joe
mikemooney
11 Posts
0
May 7th, 2011 02:00
With the configuration you have called out, if you suffer a switch failure you can maintain access to the EQL group albeit with degraded performance. While this configuration does allow you to suffer a single switch failure while maintaining access with degraded performance...
A query I would have is that would it not be better to have all connections for each controller plugged into a dedicated switch, then if it fails, let the controller recognise that ports have failed using Link Status detection and then fail over to the 2nd controller which will have 4 active connections, therefore maintaining performance....?
Thanks
Mike..
Joe S586
7 Technologist
•
729 Posts
0
May 10th, 2011 06:00
Future product strategy can't be discussed in an open forum. You're best option, is to contact your sales folks, get an NDA (Non Disclosure Agreement) in place and they will be happy to discuss some of the features on the product roadmap.
Regards,
Joe
mikemooney
11 Posts
0
May 10th, 2011 07:00
Thanks for your response, sent you a PM......
Thanks,
Mike..
bennice2002
27 Posts
0
May 17th, 2011 11:00
RenoMSP
1 Message
0
June 20th, 2012 09:00
Agreed on the link failure detection and would love to see this feature! Using two switches seems like a huge waste of bandwidth. Spanning tree will block the links on one of the two switches until a failure happens.
Chris Owens
1 Message
0
April 18th, 2014 10:00
Is this true? I agree with the OP that network link detection-based failover would be an important feature enhancement to the EqualLogic series, but I do not see any cabling guidance for the PS4100/6100 series that differs from the PS4000/6000 series. The 10GbE has its own considerations because of the active/passive nature of the 10GbE NICs, but it looks like the cabling diagrams for a PS4100 still shows a "mesh" approach as best practice.
Cabling The Dell EqualLogic PS4000/PS4100/PS6010/PS6510 Storage Arrays
1. Connect a network cable from the network switch 0 to Ethernet 0 on the control module 1.
2. Connect a network cable from the network switch 0 to Ethernet 0 on the control module 0.
3. Connect a network cable from the network switch 1 to Ethernet 1 on the control module 1.
4. Connect a network cable from the network switch 1 to Ethernet 1 on the control module 0.
5. Repeat steps 1 to 4 to connect the additional Dell EqualLogic PS4000/PS4100/PS6010/PS6510 storage array(s) to the
iSCSI switches.
Cabling The Dell EqualLogic PS6000/PS6100/PS6500 Storage Arrays
1. Connect a network cable from the network switch 0 to Ethernet 0 on the control module 1.
2. Connect a network cable from the network switch 0 to Ethernet 1 on the control module 1.
3. Connect a network cable from the network switch 1 to Ethernet 2 on the control module 1.
4. Connect a network cable from the network switch 1 to Ethernet 3 on the control module 1.
5. Connect a network cable from the network switch 1 to Ethernet 0 on the control module 0.
6. Connect a network cable from the network switch 1 to Ethernet 1 on the control module 0.
7. Connect a network cable from the network switch 0 to Ethernet 2 on the control module 0.
8. Connect a network cable from the network switch 0 to Ethernet 3 on the control module 0.
9. Repeat steps 1 to 8 to connect the additional Dell EqualLogic PS6000/PS6100/PS6500 storage array(s) to the iSCSI
switches.
If this feature is available in these new models, can someone point me in the right direction for the documentation?
Thanks.