[Most Recent Entries]
Below are the 17 most recent journal entries recorded in
Got SAN? : Storage Area Network Discussion's LiveJournal:
|Monday, November 5th, 2007|
|Tuesday, September 18th, 2007|
|Wednesday, August 22nd, 2007|
Dell introduces iSCSI frame...
When I teach enterprise-level storage classes, there's usually a snicker at the mention of iSCSI (SCSI-3 over IP networking). The reasons for the snickers is, of course, all of the folks I teach use fibre-channel to transmit data over Storage Area Networking.
Not so fast, I'll say. Even though absolutely nobody I teach messes with iSCSI in a production environment, it's coming. And when Dell gets serious about something, expect a paradigm shift:
Dell will host a major storage product unveiling in September, and we've got our money on an iSCSI storage device the company has been hinting at for months as part of the show.
The Register has learned that Dell is set to unveil a new iSCSI system called MD3000i. The unit will reportedly run on the same physical platform as the MD3000, which is Dell's direct attached SAS device.
Keep in mind that Dell already OEMs the Clariion line from EMC as a fibre-channel solution, but even buying from Dell doesn't lower the overall investment costs of FC storage. Not only do you still have to contend with the costs of a fabric infrastructure, you have to find trained, skilled SAN admins.
And that's the biggie. My 13-year old knows enough about IP networking to hook to a shared printer in my home office and to use ftp. If he sticks with that, in five years, he'll be an 18-year old networking geek, the kind that ISPs hire cheap to do admin work in data centers. Catch is, there aren't a lot off 13-year olds who play with Brocade switches. There's all these experienced network admins whose skills can be upgraded to SAN, but not without cost. SNIA training classes run $3500-5000 per week, not counting expenses. Vendor-sponsored training is at the same level.
iSCSI shifts that dynamic radically. If the front end of the storage subsystem is ethernet rather than fibre channel, a small business' network guys are ready to go.
That means there are opportunities abound for workign with those companies. Their guys may be able to hook the storage to the network, but they still need to understand concepts such as provisioning and business continuity.
Powered by ScribeFire
(Permanent link to this entry
|Monday, July 16th, 2007|
big upgrades announced by EMC...
EMC2 is rolling out a lot of upgrades to their overall storage product line, most importantly to an enterprise-level person like me, on the Symmetrix product line:
The big deal for me is the Symmetrix DMX-4 Product line, which will make some huge leaps over the already-formidible DMX-3 line:
- 4GB/s end-to-end: The DMX-3 line had 4gig front end cards for a while now, but the FC-AL back end was 2gig. The 4s will have 4gig on the back.
- Support for SATA II drives (750s): This is big for the Symm. The DMX-3s will also support 500GB SATAs.
- Faster RAID 5 and 6: up to 30% faster. More and more Symm users are opting for RAID arrays rather than mirror pairs (the original fault-tolerant Symm configuration), so this is big.
- Faster SRDF: 33% faster SRDF/S at 100km.
- Faster TimeFinder: 10x faster local replication (because of the new back end)
Additionally, there will be a several feature enhancements related to SRDF in the new microcode released for the DMX-4s that will also be available as upgrades for DMX-3s.
EMC Control Center (ECC) is also upgrading this summer as well, from 5.2 to 6.0.
El Reg has the details on the rest of the product line in the article, but the Symm stuff is near and dear to my heart, since I teach TimeFinder, SRDF, and Control Center.
(Permanent link to this entry
) Current Mood: dorky
|Tuesday, June 5th, 2007|
Siemens has made a breakthrough
in using plastic rather than glass:
Siemens researchers have demonstrated a data rate of 1Gbit/s over plastic optical fibre, a speed ten times higher than is possible with current products.
Sebastian Randel, the Siemens project manager, said the team sent an IPTV signal at 1008Mbit/s over a 100m connection in the lab, without errors or any flickering on the TV screen.
Anyone who's worked with glass fibre knows it's difficult to handle, and even more difficult to make custom lengths. I took Siecor fibre training some years back, and it took all day for me to accomplish a simple fibre splice. Plastic Optical Fibre (POF) will improve this dramatically:
POF's advantages include flexibility, robustness and ease of connection, all of which make it easier and therefore cheaper to handle - it can even be installed by us cack-handed end users. In particular, the optical core is much thicker than in glass fibre, so it is more tolerant of misaligned connectors, and it can be cut satisfactorily with a metal cutter, whereas glass needs a diamond cutter.
not to mention that plastic will (hopefully) reduce costs in the long run.
|Monday, January 29th, 2007|
Brocade just closed acquisition with McData
So there we go , no more McData..
from the ASNP mailing list today :
Dear SAN Professional,
Today we completed our acquisition of McDATA and I wanted to reaffirm Brocade’s commitment to your ongoing success.
" Current Mood: okay
|Tuesday, October 3rd, 2006|
my guess is that the FTC has no idea what either company sells:
FTC pokes stick at Brocade-McData buy
Current Mood: creative
The Federal Trade Commission has asked Brocade to give it more information over its proposed acquisition of enterprise storage rival McData.
The FTC made its request under the Hart-Scott-Rodino Act to examine the deal for potential anti-trust implications, but this looks pretty much par for the takeover course.
Brocade today confirmed its expectation that the deal will close as early as its first quarter of fiscal year 2007, ending January 27, 2007.
The company announced its intention to buy McData, in a $713m all-stock deal, in August. It intends to wield the axe post merger, looking to shave $100m in annual costs. ®
|Friday, September 29th, 2006|
teaching a class this week in Santa Clara, a guy from Cisco's FC switch group said they refer to the Brocade/McData merger as "BrokeData."
:-) Current Mood: amused
|Tuesday, August 8th, 2006|
Brocade to acquire McDATA!
just got an broadcasting email from someone in Brocade: Current Mood: surprised
Brocade Communications Systems, Inc. to Acquire McDATA Corporation
-- Transaction Expected to Be Accretive With Annual Synergies of
Approximately $100 Million
I haven't quite fully assessed the implication of this yet, since I haven't got the chance to play with McDATA switches much. at very least I hope this will revive the future of Brocade, since it's one less major competitor in the show. The only other contender I can think of is QLogic. QLogic's switches are mostly entry level ones so they shouldn't be a major threat. Meanwhile, Brocade's stock price has been in the toilet since 2002 or so, perhaps it's time to buy ?
|Tuesday, July 11th, 2006|
benchmarks for raids
I recently got an EMC AX100 RAID for work, and boy this thing is dog slow...
yeah it is a beginner RAID with SATA, but I can't believe it's running slower than single SATA HDD at times in either RAID5 or RAID1/0 mode ;(
can any of you gurus suggest a site that gives unbiased benchmark results for storage/raid ?
Thanks :) Current Mood: exanimate
|Friday, February 17th, 2006|
iSCSI on Linux
I am about to be jumping into a pilot to test software and hardware solutions for iSCSI on Redhat AS and SLES (Suse). I am trying to get a feel for the current state iSCSI on linux.
Does anyone have any experience using iSCSI with either distro? Good or bad? If so I would be interested hearing about it.
|Wednesday, January 25th, 2006|
I've got several Brocade Silkworms both directors (M14) and switches(F32). What is the purpose of the name server. I've not set that up and don't know why I'd need it. Current Mood: curious
|Saturday, November 5th, 2005|
having worked for so long on Tru64-based TruCluster systems, I'm fascinated by the OpenSSI Cluster
project. SSI="Single System Image," similar to the TruCluster's single-root concept. This needs reliable shared storage, making an FC-based SAN perfect. Thing is, even an old Brocade switch, cards, and cabling are more than I want to spend for a home-office hacking project, and I really don't want to hassle with an old-fashioned shared-SCSI SAN, a la the old StorageWorks stuff we used in the 90s.
I'm wondering if anyone has any experience with shared-USB hubs? Does anyone make a true shared USB hub, not a multi-computer USB switch? If I could build a USB-based SAN for a couple of hundred bucks, that and a couple of 1U or 2U <1ghz servers would make a wicked low-rent cluster... Current Mood: curious
|Monday, October 31st, 2005|
SAN Horror stories
I've worked on some really bad units. Even some current SANs have awful interfaces. What are some of your SAN horror stories, either with just awful setups or really bad "uh-ohs" ?
|Saturday, October 29th, 2005|
Differences between SAN and NAS?
I'm used to dealing with small(ish) networks where data that needs to be accessed by multiple machines on the network is stored on one or two fileservers.
Is it correct that one of these fileservers represents an example of NAS? (Network Attached Storage) Or is NAS a separate class of device entirely?
I went through the introductiory section of DotHill's SAN tutorial ( http://www.dothill.com/tutorial/
) and for the most part, it looks like a SAN is essentially a high-speed LAN consisting of nothing but fileservers... and routers, via which LANS & WANS connect to access the data.
Is this description appropriate? Or is there something I'm missing?
|Friday, October 28th, 2005|
I created this community for all the SAN
professionals out there. ;) Whether you just want to learn more about Storage Area Networks or are a SAN admin, feel free to ask questions, discuss current events, product releases or anything SAN related. If you have any questions don't hesitate to contact me.