#Looking at ASAr2 ... Feedback?

1 messages · Page 1 of 1 (latest)

civic wing
#

Any current users able to provide feedback on your experience so far?

  • general likes and dislikes
  • successes and challenges
  • feature wish lists
earnest geyser
#

I have two ASA A20s and one A90. From box to provisioned in 90 minutes.
The storage availability zones sold me on the R2s. Not having to balance volumes and aggregates between nodes makes it more like other block arrays.
The major dislike is the number of bugs especially on the A20. Right now every 100 or so days one of the nodes will stop reading slot temperatures correctly. The node management becomes extremely slow and the current fix is to takeover/giveback.
Snapmirror to and from non ASA R2s is needed.
Support is better than FAS/AFFs. Tier 1 guys don’t know anything about them so you don’t have to deal with them 😃

#

Temperature bug: CONTAP-490067

haughty copper
#

the fact that all paths are active and up is especially neat for systems which do not handle even short "all paths down" events gracefully (looking at you, AIX, but there are others). That's a big win. Also, SnapMirror activa sync with active paths on 2 sites is cool.
My only gripe is that it's incompatible with FAS/AFF systems when it comes to SnapMirror/SnapVault for backup. I hope that'll be "fixed" soon so that we can use our existing Nearstores as backup targets

gray hemlock
#

Data Reduction Rates are very hard to see and see where the numbers are coming from. You need to dig into CLI Commands.
It’s like „4.5:1“, the LUN is using 12TB, in vSphere you see 18TB consumed. Now good look to find out where 4.5:1 comes from and how much storage is actually used by this LUN.
Storage capacity management is very hard due to that reason.
However, the worst about ASAr2 is the way NetApp Tools support it. NAbox (harvest), AIQUM, Config Advisor etc., they all don’t support r2. I am fine with this for something very fresh to the market, but it’s been there long enough.
I also dislike NetApp‘s silence when it comes to the hard truth about the massive performance impact of solutions like SM-AS (this applies to non-r2 as well). It’s not even in fusion…
We have also discovered that for ethernet protocols the operating systems timeout values for paths being unreachable is not addressed at all. No recommendations, no settings in OTV, no feedback, no communication whatsoever. In the current state, it is just as good as any other ethernet based SAN storage in terms of path failover.
It’s also funny to see NetApp violating their own best practices in ASA r2, for example when it comes to the raid groups. It’s not a big deal, but funny.
As already mentioned, one of the best things is that you don’t have to care about different aggregates anymore.

light venture
gray hemlock
# light venture Yea the NetApp Tools support is not the best, even though the most important one...

We have done performance tests with NVMe/TCP and iSCSI. With SM-AS active, you have ~40% less IOPS if I recall it correctly. Once one side goes down, your IOPS go up to a level of a single HA system. Once the failed side comes back IOPS drop again.
I don’t know if the impact is still that massive. We did most of the test back in February or so. There haven’t been any official comment on it as I am aware, not even statements it has improved.
We have tested it with 2x ASA A70, they have been in the same rack and all the connectivity as well. So distance / latency wasn’t a factor.

Thanks for the update on NAbox, haven’t checked it recently but I will in this case

elfin herald
#

Good and interesting thread! Is the drop equal for both read and write iops?

gray hemlock
haughty copper
#

we saw a huge drop (40-50%) in early EVP and RC versions too, but in recent 9.16.1 versions the drop is only around 15-25% from what we saw

civic wing
#

Thanks for all the feedback!

I'd noticed the "tools" issue while digging around. Seemed to be a lot of push for DII (which costs us more $$) and Harvest (while i like it, it is not a supported tool like AIQUM and can be a challenge in our environment)

Right now we're strictly looking at FC environments... heavy on the virtualization side (VMware and KVM) and some generic block storage to host scenarios

civic wing
steady ingot
civic wing
# steady ingot Re the DII cost point - I take it you don't see the capabilities (FC topology, c...

To be clear, I like DII and see value in it, but it has a cost.

It just feels awkward when you have an 'included' off-box solution for AFF/FAS (AIQUM) but nothing for ASAr2 ... sure, there are work arounds, but I have to treat it as a special case vs other NetApp deployments in the environment

Competing solutions still have that supported off-box product as part the offering:

  • Pure - Pure1
  • HPE 3PAR/Primera - SSMC
  • Dell PowerMax / PowerStore - Unisphere
earnest geyser
gray hemlock
#

But this is probably due to having 32 disks only in the system and the way the population rules are. As a full blown 48 disks system I expect to have 24+23 or something similar

stiff iris
#

As @light venture mentioned, Harvest supports collecting metrics from ASAr2. We've been able to collect metrics since the 24.11.1 release.

A few weeks ago we added storage-units and availability-zone templates too. The default Harvest dashboards show the available ASAr2 metrics. Recently, we've discussed adding a separate ASAr2 dashboard since the standard set of dashboards include more details than ASAr2 needs. Do folks think that would be useful? If so, please +1 the current issue https://github.com/NetApp/harvest/issues/3534

sinful viper
#

Does someone know when MetroCluster will be supported on ASA R2 systems ?

sinful viper
#

there is nothing about MetroCluster support in HWU and NetApp docs for ASA A50/70/90 and A1K

haughty copper
#

MetroCluster for ASA r2 is not planned. SMas is supposed to replace that functionality

sinful viper
#

SMas states for SnapMirro async/sync ?

haughty copper
#

SnapMirror active sync

sinful viper
#

thanks a lot @haughty copper

stiff iris
#

The latest nightly build of Harvest now includes an "ASAr2 Overview" dashboard that shows ASAr2 storage units and hosts as shown below. We appreciate any feedback folks have. Thanks!

image.png
civic wing
#

if we land an ASAr2, I'll certainly be checking it out!

curious, are their views that show performance metrics of a combined set of storage units? I guess this could be by members of a consistency group. Best comparison would be a "Storage Group" on a PowerMax

stiff iris
#

hi @civic wing yes, the performance metrics of combined storage units are shown in separate panels. Right now, those panels are in the lun or volume dashboards instead of the ASAr2 one. Maybe it makes sense to pull some of those into this Overview dashboard. Not sure yet, we're trying to avoid duplicating panels that exist in other dashboards