EVA8000 vs. EVA8100

[Fedon Kadifeli]

I heard that the EVAx000 series products were discontinued and the new versions (EVAx100) were introduced. I searched HP's web site to find out what are the differences, e.g., between EVA8000 and EVA8100, but I could find enough information.

Can anyone point out the main differences between EVA8100 and EVA8000?

Thank you!

[Rob Leadbeater]

Hi,

As I understand it, the main differences are that the controllers, switches, loops etc. have been upgraded to 4Gb/s.

There are also performance improvements due to faster processors in the controllers.

Hope this helps,

Regards,

Rob

[Uwe Zessin]

Most important thing is the improved fault-isolation in the disk drive enclosure. The claimed performance increase can only be seen in special, large configurations.

[Fedon Kadifeli]

Can you elaborate more on "the improved fault-isolation in the disk drive enclosure"?

Thank you!

[Uwe Zessin · Accepted solution]

Sure, the old versions basically had an Arbitrated Loop HUB on the A and B loops. Now, hubs do have bypass logic, but can get confused by bad-behaving disk drives. I've been told that the controllers can tell the hub to isolate(bypass) a disk drive via the extra CAB cable, but that does not always seem to work.

The new enclosures have little loop switches which can send the traffic directly to the desired disk drive - the frame no longer has to pass all disk drives in the enclosure. The loop switch establishes individual 'looplets' to each disk drive within the enclosure - similar to the external loop switches between the controllers and the disk drive enclosures. The embedded loop switch is supposed to provide a better fault isolation within the enclosure.

We will see in the next months how good it works ;-)

Don't get me wrong. Usually, the EVA is a very reliable box, but it is not 100% perfect.

[Del_3]

The 8100 is supposed to be 25% faster.

[Uwe Zessin]

Sure, if you use one of those wonderful benchmark programs that create a throughput of 2450 MegaBytes per/second or 225,000 IOPS ;-)

In a fully equipped 8x00 it would mean 225,000/240 = 935.5 IOPS per disk if the benchmark were not carefully written to stay on the controllers cache. Too bad that most customer data is too large for the cache.

[Mark...]

Hi,
Diffs in a nutshell-
HSV200-B =41xx/61xx ctrl
HSV210-B = 81xx ctrl
Supposed to be 25% faster [new ctrl CPU]
200-B = 2Gb cache per ctrl [up from 1Gb]
210-B = 4GB cache per ctrl [up from 2Gb]
New fully switched backend loop switches still 2Gb
New point to point shelves M5314C, still same disks & emu - new I/O modules
Improved fault tolerence / troubleshooting [backend loops]
Total disk capacity = same as older 4/6/8
4Gb host ports = same as before
XCS6.110 for 41xx/61xx/81xx
Command View 7
New LUN shrink capability
RSM V3.0
Think thats it.

Mark...

[Del_3]

You can tell Uwe doesnt work in sales.

[Uwe Zessin]

Right, but I do pre-sales and implementation.