The Fox in IBM’s Storage Henhouse

Posted on Dec 4, 2009 in Records Management News

http://www.itjungle.com/tfh/tfh113009-story04.html

Mad Dog 21/21: The Fox in IBM’s Stor­age Henhouse

Pub­lished: Novem­ber 30, 2009

by Hesh Wiener

Moshe Yanai became suc­cess­ful by tak­ing enter­prise stor­age busi­ness away from IBM. He led the team that cre­ated the EMC Sym­metrix, which became the lead­ing stor­age prod­uct at IBM’s glass house accounts. EMC and Yanai parted ways in 2001 and after a decent inter­val Yanai founded a new stor­age ven­ture, XIV (pro­nounced Ex Eye Vee). IBM acquired XIV at the start of 2008, nam­ing Yanai an IBM Fel­low. Yanai may be able to clob­ber EMC for IBM, but to suc­ceed he will also have to kill off IBM’s flag­ship DS8000 array with his XIV boxes.

The rea­son XIV just might be rev­o­lu­tion­ary, at least by IBM stan­dards, is that it is a disk array that has some of the attrac­tive char­ac­ter­is­tics of just about every kind of array in the alpha­bet soup of today’s stor­age indus­try. Basi­cally, an XIV box has a front-end based on X64/Linux servers with soft­ware and inter­faces that let the machine talk over Fibre Chan­nel, Eth­er­net, iSCSI (which uses Eth­er­net), and if there were a need, any other fast hookup the mar­ket might want. Behind the gate­way machines sit a pile of X64 servers that talk to large SATA disk dri­ves. Today the dri­ves are 1 TB each; tomor­row they will be 2 TB and maybe in 2010 or 2011 they will go to 4 TB. The hard­ware com­po­nents are stan­dard and cheap. Each XIV box can have up to 180 dri­ves; this com­ple­ment of disks yields 79 TB of stor­age capac­ity after deduc­tions for inter­nal mir­ror­ing and spares. Total cache in each box can be up to 120 GB. (Details galore for the XIV stor­age clus­ter can be found in IBM’s XIV Red­book.) But as all the hard­ware is based on indus­try stan­dard tech­nol­ogy, it is the soft­ware that really adds value.

Moshe Yanai Calls The Tune: For­merly the lead devel­oper for the EMC Sym­metrix that killed IBM’s pre-array stor­age busi­ness, today he is an IBM Fel­low whose XIV disk array could bump off older high-end prod­ucts from every ven­dor, includ­ing Big Blue.

The XIV soft­ware dupli­cates data (mak­ing the machine RAID 1) and scat­ters it across the full set of dri­ves in the cab­i­net. This, says IBM, lets the box rebuild a failed drive quickly (20 to 30 min­utes) and accu­rately. Data scat­ter­ing makes it very unlikely that a whole file or dataset gets clipped if a drive fails, but a two-drive fail­ure, which is a very unlikely event, can cause a whole box to lose via­bil­ity. There has been a blog rant about this but peo­ple who read through the mate­r­ial will most likely end up trust­ing the disk sub­sys­tems more than the bloggers.

The actual machine is not quite this sim­ple but nei­ther is it as elab­o­rate as the DS8000 boxes IBM sells to its high-end server cus­tomers, includ­ing main­frame and Power Sys­tems shops and includ­ing AIX 6.1, Linux, and i 6.1 (the lat­ter through the Vir­tual I/O Server par­ti­tion) on the Power boxes. IBM says the XIV is a bar­gain and that it also is less power hun­gry than alter­na­tives. While IBM’s sales­force makes these claims when com­par­ing the XIV to non-IBM alter­na­tives, the very same com­par­isons can be made to the DS8000, which remains in the spot­light in part because it is the only large IBM disk array that z/OS will talk to.

There may be a per­for­mance case for the DS8000 com­pared to the XIV (and other high-end arrays), at least for cer­tain kinds of work­loads, but in prac­tice there are no inde­pen­dent bench­mark tests for big stor­age sub­sys­tems. Indus­try folk­lore sug­gests that a box like the DS8000 should shame the XIV in trans­ac­tion pro­cess­ing appli­ca­tions, while the XIV would win when data is not struc­tured. Unstruc­tured (or incon­sis­tently struc­tured) data shows up a lot in pop cul­ture sites like Face­book but also is the nature of med­ical records sys­tems (with com­puter files, doc­u­ments, pho­tos, videos, scans, etc.) and insur­ance claims data­bases, to name only two of many appli­ca­tions where large disk arrays are used.

Regard­less of widely held pre­con­cep­tions, how­ever, the XIV may be bet­ter than DS8000 advo­cates (such as users who have a big invest­ment in the machines and some IBMers with careers tied to the prod­uct) seem to believe. Bank Leumi uses XIVs for every­thing, includ­ing trans­ac­tion pro­cess­ing, and says it is get­ting great results. Recently, IBM has been hold­ing dog-and-pony shows for users and ana­lysts that fea­ture a grow­ing num­ber of user orga­ni­za­tions that have bet their strate­gic stor­age on XIV equip­ment. \r\nFor now, IBM is pro­tect­ing its DS8000 base by not sup­port­ing XIV under z/OS. If you want to use a XIV on a main­frame, you have to hook it up using zLinux. (On Power-based servers, AIX and Linux can talk directly to the XIV arrays, but i 6.1 has to work through VIOS to reach the machine.) Clearly, direct links between Power Sys­tems or main­frame machines and XIV clus­ters could hap­pen just as soon as IBM is will­ing to let XIVs can­ni­bal­ize the DS8000 base. Actu­ally, the deci­sion would not so much be IBM’s but that of high-end server shops, who will flee the DS8000 the minute they fig­ure out that XIV is going to be the future of IBM enter­prise stor­age, a process that is begin­ning to get underway.

Disks at Midyear: This IDC snap­shot of the high-end stor­age mar­ket dur­ing the sec­ond quar­ter of 2009 shows just how hard it is for IBM to rise to the top. It may be sec­ond after EMC, but it had only 15 per­cent of indus­try sales.

Out­side the main­frame base, there’s plenty of sup­port for XIV. But for IBM, which is not the prime ven­dor to the Web 2.0 world of Google, Ama­zon, eBay, and their ilk, the strate­gic bat­tle is the fight to store glass house data. And in that world, IBM has to grow by 50 per­cent to catch up to EMC.

In the sec­ond quar­ter of this year, which is the most recent quar­ter for which IDC has issued a pub­lic report, EMC had 21.5 per­cent of the mar­ket IDC calls exter­nal disk sys­tems. IBM had 14.9 per­cent of this $4.1 bil­lion (in the quar­ter) seg­ment. Hewlett-Packard fol­lowed with 11.4 per­cent, Dell had 9.89 per­cent and NetApp had 8.9 per­cent. Gart­ner says more or less the same thing about the mar­ket, peg­ging EMC’s share of what it calls con­troller based disk stor­age at 21.8 per­cent in this year’s first quar­ter, IBM at 15.7 per­cent, HP at 10.3 per­cent, Dell at 9.2 per­cent, Hitachi (which also sells sub­sys­tems through HP and Sun) at 8.7 per­cent, and NetApp at 8.5 percent.

The 2009 mar­ket data describes a seg­ment that’s headed for a year-on-year decline of 15 to 20 per­cent com­pared to 2008, which itself was not glo­ri­ous. IBM’s main­frame sales data con­firm times are tough in glass house coun­try, and with banks dying like mayflies, the out­come for the indus­try, for each ven­dor and for var­i­ous prod­ucts is quite hard to pre­dict. Nev­er­the­less, it’s prob­a­bly safe to say a story about cheaper stor­age will get users’ ears.

It’s very hard to tell just how cheap a XIV is com­pared to alter­na­tives, in part because IBM report­edly pumped out so many eval­u­a­tion machines, mean­ing free­bies, to seed the mar­ket. Now EMC blog­gers are not about to praise the XIV but they also are not going to put com­plete balder­dash on their Webs because it’s hard enough for them to get eye­balls under the best of con­di­tions. But IBM fans (and peo­ple who like to gripe about EMC) might want to dis­count Big Blue blog­gers at least as much as those on the EMC pro­pa­ganda squad. One of IBM’s (until recently, any­way) vis­i­ble and often enter­tain­ing blog­sters, Tony Pear­son, seems to have just dis­ap­peared … or what­ever you call it when a whole lot of Google point­ers that for­merly worked now take you to 404 country.

It’s pos­si­ble that these devel­op­ments in the murky world of stor­age mar­ket­ing and per­sua­sion stem from the heated dis­cus­sions about the actual cost and real green­ness of XIV. The dis­pute, which we found close to impen­e­tra­ble, stems from the fact that a XIV box is at its most cost-effective and power friendly when it is fully con­fig­ured. In fact, it may be the case that IBM builds every XIV fully loaded, the way it builds large servers with a full load of engines, and then turns on pieces of the machine based on how much a user pays.

IBM’s XIV Disk Array: A box con­tain­ing X64 servers, big and slow SATA disks, stan­dard inter­con­nec­tion tech­nol­ogy and soft­ware that turns it all into a daunt­ing stor­age product.

It turns out that a seri­ous prospect for the XIV or an alter­na­tive from IBM or one of its rivals can get good answers to hard ques­tions quite directly. First, get IBM and any other ven­dor in the run­ning to quote a five-year total cost includ­ing main­te­nance, soft­ware to pro­vide all the required disk array func­tion­al­ity, etc., includ­ing mir­ror sys­tems and what­ever it takes to do mir­ror­ing if the mir­ror is going to be remote. At the same time ask the sales rep for power and heat data for the machin­ery as it will actu­ally be con­fig­ured. Any­thing less is just ask­ing for ugly surprises.

Even when you have done that, you may not have all the answers unless you give ven­dors’ financ­ing arms (and the hand­ful of inde­pen­dents whose names come up if you search for “used IBM main­frame stor­age”) a chance to offer sec­ond­hand equip­ment. Just as you should in the case of new gear, you have to press the ven­dors of used disk arrays to quote total costs includ­ing main­te­nance and func­tion­al­ity soft­ware as well as power and heat.

Also, if you think you might be mov­ing some or all of your work from one plat­form to another dur­ing the next few years, you need to get prices that include all the rel­e­vant inter­fac­ing hard­ware and soft­ware. While it ought to be suf­fi­cient to spec­ify a stan­dard such as Fibre Chan­nel or iSCSI (over Eth­er­net), be spe­cific. In this game, you can’t count on good surprises.

The result might cause a bit of a stir. On just about any mea­sure where a XIV looks good com­pared to, say, a Sym­metrix or Sun Microsys­tems Stor­ageTek array, it is going to look great com­pared to a DS8000, if IBM’s flag­ship disk sub­sys­tem is in the running.

It’s not that IBM fel­low Yanai wants the DS8000 to tank. It’s that he prob­a­bly doesn’t have any choice if he wants his XIV device to emerge a winner.

Read more here          http://www.itjungle.com/tfh/tfh113009-story04.html

Com­pli­ments of File­Man Research

Cary

_____________________________________________________

Cary F. McGov­ern, CRM

File­Man