Sun Availability Suite blog
Dale Ghent | January 30, 2007The maintainer(s) of Availability Suite (aka AVS) now have their own blog. I can’t wait to learn more about it!
Technorati Tags: solaris, opensolaris, storage, sun, avs
The maintainer(s) of Availability Suite (aka AVS) now have their own blog. I can’t wait to learn more about it!
Technorati Tags: solaris, opensolaris, storage, sun, avs
The OpenSolaris community got a huge present today by way of Sun’s open sourcing of its StorageTek Availability Suite. This announcement by Jim Dunham goes into the details, but here’s a summary:
Availability Suite is comprised of two primary components:
I applaud Sun for releasing this! OpenSolaris now has far more and robust storage tools than any other FOSS (or otherwise, for that matter) OS out there.
Hey Sun, how about releasing ESM AA next, eh? There’s Aperi, but ESM AA looks to be far more mature.
Technorati Tags: opensolaris, solaris, sun, storage
Sun and Intel announced a new partnership between the two companies today, with both CEOs presenting at a news conference this morning.
Sun hasn’t had a Intel CPU in its product line since Sun discontinued its Pentium 4-based V60z server several years ago when the company was teething its new x86 product line. This product line eventually developed into the exclusively AMD Opteron-based servers we have today. Intel/Xeon was out, AMD/Opteron was in. With today’s announcement, both Intel and AMD will now share Sun’s x86 product portfolio.
I’ve noticed that reaction to this news has been mixed, with some saying it’s good, and others saying “WTF, mate?”. Sun’s Opteron-based Galaxy servers are top-notch, so this has lots of people utterly surprised… like having a great night out with someone and then being dumped on the doorstep.
Well, it’s surprising news to say the least. As I thought about it more, though, it isn’t bad at all for Sun, and really isn’t all that forboding for AMD. In exchange for a Intel-based product line, Intel will seriously push Solaris for Sun. That is exquisitely good news. Sun now has a product line which can serve both AMD and Intel customer preferences.
Think of that “iPod Halo Effect”. Sun doesn’t have to turn away customers who want Intel CPUs now, and with Intel pushing Solaris, hopefully more applications and thus more Solaris installations will be in customers’ data centers. I dare say that those customer will like Solaris, and perhaps look to buy (more) Sun servers.
Besides, Sun isn’t the only company to straddle the divide between AMD and Intel. Dell, a traditionally staunch Intel ally, added Opteron servers to its product line last year, as did IBM. HP has offered systems with CPUs from both companies for at least 1.5 years, if not longer.
It’s a move that makes sense, especially for Sun, and that in itself something we should applaud… moves that make sense (duh!)
Solaris 10 11/06 was released in late 2006 with a plethora of new features, and among them a new tool called mpathadm, which comes as part of the SUNWmpathadm package.
Before I delve into how this specific tool works and how it helps when managing multipathed storage in Solaris, I’ll give some background on what multipathing is and how it is implemented in Solaris.
Read the rest of this entry »
This month there has a been a noticeable increase in buzz from IT pundits discussing whether Solaris now has the feature set critical mass required to win over exising Linux users and shops.
On one hand, I think it’s tres cool that Solaris is and has been getting lots of positive press after a long time (many years) of being shunned as “grandpa’s UNIX” or even outright ignored. On the other, I have to wonder if these pundits are getting a little too “rah rah ree” vis a vis this new rising star of the OS popularity contests. As nice as it is, this kind of speculation has been going on for some time now and I have never seen any hard numbers to indicate whether such a shift is afoot or if it has even happened (yet).
This isn’t to say that Solaris’s newfound popularity (some would say it’s a revival) is without technical merit. The press and technologist community have been crowing about big new features such as ZFS, Zones, and DTrace for some time now. I’m worried, though, that the non-Solaris user who is being courted will get burned over on these specific and oft-repeated messages and come to think that those “Big 3″ features are the only things that separate the Solaris of today from the derided Solaris of even 4 years ago.
While I and my fellow Solaris admins know that there are a whole host of other enhancements and features in today’s Solaris, I don’t think that these are being adequately enumerated and explained in generic terms to the non-Solaris admins out there. Not everyone is looking for a new file system. Not everyone is looking for a tracer of the type DTrace is… but I bet there are a lot of Linux admins out there who are trying to do something that Linux doesn’t or doesn’t do well, and Solaris does. The “killer” feature that could win someone over could be as mundane as resource management or Solaris’s excellent mass storage management. Jörg Moellenkamp has similar thoughts on this subject as well.
Who knows? All I know is that we can’t continue to harp on the same stuff, or people will get the wrong impression that Solars is just those things. It’s far more than that, and we have to clearly promote those less sexy but still awesome aspects.
Technorati Tags: solaris, opensolaris, linux, software
It’s not often that I’m so utterly disappointed by a product that I feel the need to write about my experience with it, but a situation at work with two newly-arrived Sun StorageTek 6140 disk arrays has cetrainly enraged me enough, and boy do I feel the need to rail against vendors who cripple their products on purpose.
NOTE: As of 8 Feb this issue has been resolved, but read on if you want to head about the saga.
Read the rest of this entry »
Sun released the third maintenance update for Solaris 10 yesterday. The What’s New documentation covers all the features it adds to Solaris 10. There’s so much Good Stuff in this update that’s it really should be called “Solaris 10.5″. The ones most exciting and immediately useful to me:
Basically, this adds the ability to manage Solaris’s fibre channel multipathing without having to manually edit /kernel/drv/scsi_vhci.conf and rebooting. It’ll also present you with clear status on multipath’d links without having to piece together lines from syslog. See the new mpathadm(1m) command.
Logical hardware partitions for sun4v-based servers! This means you can now take advantage of the hardware hypervisor on your Sun Fire T1000 or T2000 to carve it up into up 32 independent logical domains, each running their own OS instance.
Recursive ZFS snapshots is a godsend to those (like me) who have many nested ZFS file systems. One command will now snapshot them all, rather than having to explicitely snapshot each one individually. DP RAID-Z (raidz2) is another feature comperable to what’s known as “RAID 6″, and of course hot spare device support, the usefulness of which is self-evident.
An article over at Computer World reveals a rumor that Google might be investigating OpenSolaris as its future primary platform. The article goes on to describe the OpenSolaris community as it stands today, just over a year after its release. I’m quoted at the top of page 3.
Technorati Tags: opensolaris, solaris, google
I went out today and picked up a 1x PCIe eSATA card made by SIIG (model SC-SAE412-S1) and put it in my OpenSolaris dev box. As expected, the si3124 driver picked up on the Silicon Image 3132 chip on this card and was fine with it. Now I just need to get a multi-bay external SATA disk box to hook up to this card so I may continue with my NAS appliance project.
It seems that the non-RAID SATA/eSATA cards made by SIIG are all based on either the SI 3124 (for PCI/PCIX) and 3132 (for PCIe) chips, so they’re a safe bet if you’re looking for such a thing for your Solaris 10/OpenSolaris box, as the si3124 driver (part of the SATA framework) should jive with it.
cfgadm -al output with a Maxtor STM3500630AS disk attached to each of the two eSATA ports on this card:
Ap_Id Type Receptacle Occupant Condition sata0/0::dsk/c7t0d0 disk connected configured ok sata0/1::dsk/c7t1d0 disk connected configured ok
/usr/X11/bin/scanpci output:
pci bus 0x0002 cardnum 0x00 function 0x00: vendor 0x1095 device 0x3132 Silicon Image, Inc. SiI 3132 Serial ATA Raid II Controller
Those two disks are now a mirrored set in a ZFS pool:
[daleg@helium]~$ pfexec zpool status local
pool: local
state: ONLINE
scrub: none requested
config:
NAME STATE READ WRITE CKSUM
local ONLINE 0 0 0
mirror ONLINE 0 0 0
c7t0d0 ONLINE 0 0 0
c7t1d0 ONLINE 0 0 0
errors: No known data errors
It works!
Technorati Tags: opensolaris, opensolaris+qube
I read over the specs of Sun’s new mid-range storage models, the StorageTek 6140 and 6540, which were announced yesterday (8/11/2006) and the corners of my mouth must have visibly sank after scanning through the specs and not seeing the one big feature I was hoping to see in any new arrays from Sun. This feature is iSCSI.
Don’t worry, I checked and double-checked, and iSCSI is not a feature. With there being only 100Mb-capable ethernet ports on the controllers and not 1Gb, there’s also no holding out hope that it could be a feature-add at some point in the future.
Unless I’m overlooking some extraordinary feature that these two new arrays introduce, these are just re-skinned logical upgades of what Sun has always offered. 4Gb Fibre Channel and 4GB of cache. Higher IOPs. In one respect, that’s all nice and stuff, but it’s not like anyone could claim that they didn’t see that coming.
Technorati Tags: Sun, storage, Sun Microsystems, Solaris, RAID