The Unified Storage System FAQ

Analytics

Question:
Customers have commented that they would love to have a 'dataset' for capacity utilization. They would like to be able to view utilization 'By Project, By File System/LUN, As a Raw Statistic' over time. I understand that isn't instrumented using DTrace, but I believe it could be visualized using Analytics. It wouldn't need to be instrumented each second, but once every minute or few minutes would probably work. This feature could be used to show capacity trending over time and facilitate planning. Customers are also asking for 'Threshold Alerts' for capacity: 'If Capacity Exceeds 80% for 60 minutes ... then send an email'.

Response:
We know everyone wants this: we want it too. It didn't make our first release but we intend to add this and various other kinds of analytics in future software updates and expect this sooner rather than later.

Filesystem

Question:
Customers ask how we are protected against the file system fragmentation issues that cause performance to degrade as capacity is utilized. The follow up is if we are going to have performance problems, will we offer a defrag tool?

Response:
ZFS defrag analysis is ongoing, but we believe current ZFS is non-pathological. If we discover we could enhance performance using some type of defragmentation algorithm it would be more automatic and not some tool you have to run manually.

Quotas

Question:
Some customers have asked whether we are able to offer quotas for individual users on the system (ie home directories).

Response:
User Quotas will be part of a future software update. In the meantime you could create a separate filesystem for each user.
Update This is now available in the Q3 software release.

Question:
Why does the system allow the quota of an individual filesystem to be greater than the quota of the project that contains it.

Response:
Quotas are an intentionally abstract concept. You may be enforcing a limit on a filesystem that happens to currently be in a smaller project due to organizational reasons, for example. It's similar to allowing you to set a quota larger than the pool - just because you're limited by the pool size right now doesn't mean you will always be so limited. Mostly, there's no good reason to enforce it, and it significantly complicates the implementation.

System Maintenance and Monitoring

Question:
What data will be available by SNMP? Will the system support only have basic metrics (ie. a standard MIB) or will there be a specific NAS MIB that offers capacity or performance statistics?

Response:
The Storage 7000 series will initially include three basic MIBs: basic machine status, fault management traps and problem reporting, and projects and shares. The second two are Sun specific.

Question:
The 'Scrub' option is very nice, and I would like to schedule system scrubs automatically, can this be done?

Response:
There is an open Request For Enhancement for this. It's much more than picking a time to do a scrub, as what you actually want to do is adjust the rate of background scrubbing to ensure a regular rate of scrubbing over a period of time. Some of the pieces are in place to do this, but there's more work to be done.

Question:
In my testing, hardware alerts seem to work nicely (unplugged power cord, power supply). What I noticed was that when I pulled one of the boot drives however, it doesn't rebuild to spare because the spare is in the other pool, and it doesn't alert to the failure of the drive either. Even when we pulled a data drive, it mentioned the resilvering in the log, but didn't show a disk alert using the orange '!' icon. I would think that disk failures should alert, especially boot drive failures where there is no spare to work with.

Answer:
There are no spares for the root pool, so there is nothing to spare in. More importantly, removing a drive is not the same as faulting a drive. This is an intentional hotplug, and therefore there is no associated problem. You get alerts for drive removal, but problems indicate faulty hardware or software. That is intentional and will not change.

Question:
Where can I find the latest software updates for the Sun Storage 7000 Series?

Answer:
Software updates for the Series 700may be downloaded from the Sun Download Center. A table of past and present software releases is located here with pointers to specific releases.

Projects and Shares

Question:
While examining project based snapshots, deletion of the schedule destroyed all snapshots created by that schedule. Is that intended behavior? Is there a way to mark a snapshot for retention?

Response:
Yes, that is the intended behavior, and that is documented in the system's help wiki (see the help button at the top right of the BUI). There is no way to mark an automatic snapshot for retention.

Question:
I have an application that requires providing 'chown' privileges for any user on an NFS filesystem. Solaris has an option in /etc/default to allow this behavior and other NAS appliances have an option to enable this but I don't see any way to enable these privileges in the Sun Storage 7000 Series.

Response:
This is not possible today however this capability will be added in the future.

Networking and Protocols

Question:
Does the Sun 7000 Series support 'Network Flow Control'?

Response:
Yes, it does*
Note - this question isn't clear. I think the real question is "Do we support network QoS based on layer-2 or layer-3 policies?" The answer right now is "Not yet."

Question:
How do you configure link aggregation on the switch-side? Could you give an example?

Response:
Link Aggregation requires configuration on both the appliance and switch side of the connection. If your switch supports LACP (IEEE 802.3ad Link Aggregation Control Protocol) then both sides should have LACP link negotiation turned on. The appliance side settings for enabling LACP are defined at the datalink layer of the network configuration view. Valid values for LACP-enabled aggregations are "Active" and "Passive." The switch must be set with a compatible value for the corresponding switch aggregation. Compatible appliance/switch value combinations are Active/Active, Active/Passive, and Passive/Active. The appliance setting of "on" forces aggregation of the associate ports without using LACP. Forcing the ports to aggregate without using LACP may be necessary for compatibility with switches that either have LACP turned off or do not support 802.3ad LACP.

Example:
The appliance has a primary management interface configured on nge0 using a simple device-datalink and single static IP. Additionally, the administrator wants to configure a link aggregation using the remaining built-in devices (nge1,nge2, and nge3). These 3 ports are connected to a Cisco Catalyst 3750-series switch (ports 11, 12, and 13 respectively). The administrator chooses to use an LACP setting of "Active" for the appliance and "Passive" for the switch. Finally, the administrator chooses the load balancing policy for outbound traffic will be based on IP address, while the incoming traffic will use layer-2 MAC addresses.

On the appliance side - the administrator configures this from the browser interface (Configuration->Networking). Click Add Datalink and select the LACP option. Then select the 3 devices that will comprise the aggregation (nge1,nge2, and nge3). For Aggregation Settings - choose Mode=Active Policy=L3, and Timer=Long. Click Apply. Finally, add an IP interface and associate it with the new (aggregation) datalink.

On the switch side, issue the following configuration commands (reminder - this is a Cisco Catalyst 3750 series running IOS 12.2x)

1.) Configure the switch global setting for load balancing (layer-2)

# config term
(config)# port-channel load-balance src-mac

2.) Configure the port channel group and assign lacp settings to each member interface

(config)# int range Gi1/0/11 - 13
(config-if-range)# channel-protocol lacp
(config-if-range)# channel-group 1 mode passive

3.) Confirm your configuration and lacp state of each member link

# show etherchannel summary
Flags: D - down P - in port-channel
I - stand-alone s - suspended
H - Hot-standby (LACP only)
R - Layer3 S - Layer2
u - unsuitable for bundling
U - in use f - failed to allocate aggregator
d - default port

Number of channel-groups in use: 1
Number of aggregators: 1

Group Port-channel Protocol Ports
------+-------------+-----------+-----------------------------------------------
1 Po1(SU) LACP Gi1/0/11(P) Gi1/0/12(P) Gi1/0/13(P)

Question:
How do you configure VLANs on the switch side? Could you give an example?

Response:

Hardware

Question:
I want to add Logzillas to my 7210. What slots should they go in, and do I need to do anything special to have them recognized?

Response:
They can go in slot 3 or 11. Please be aware that you cannot add a ssd to a 7210 that already has data on it without destroying the storage pool first. This would obviously mean you would need to backup your data, destroy the pool, swap in the Logzilla SSD's, rebuild the pool, and then restore your data.

Question:
What slots do Logzilla's go in if I have a 7310/7410?

Response:
Depending on the number of LZ's to be installed in the disk expansion chassis', they would normally go into slots 4,8,12, and 16. If in a cluster and you want LZ's split between both heads, one recommendation is to use slots 4 & 12, and then 8 & 16. Then split the jbod's in half with drives 0-11 (remember 12 drives is the minimum diskset size) assigned to node a and drives 12-23 assigned to node b. You can add more LZ's in another jbod the same way. An alternative to this is to segregate the pools between the two nodes by whole jbods. In that case, one would add LZ's to slot 4 in two different jbods, then slot 8 in both jbods, and so on. Then pools would be built by placing a jbod with LZ's in each pool. The other recommendation with respect to Logzilla placement is to try and keep them as close to the front of the jbod string as possible. This would mean placing LZ's in jbod 1 across the hba's before putting them into the second jbods in the strings. This helps to lower the latency incurred going across the cables and SAS expanders.

Other FAQ's

Check out the FAQs posted by partner Corporate Technologies at http://ctistrategy.com/sun-7000-faq/