In my last post (too long ago, I know) I talked about the Weather Company acquisition. It is open for business under the IBM brand and we are incorporating Watson technology. The interesting thing is that Watson will be used to make advertising smarter. Get ready for big changes in how we interact with advertising.
Speaking of getting ready, this time around I'm ready for another 30 inches (76 cm) of snow like we had last year. The International Cub still isn't ready, as it is so heavy we need a crane to pick up the body to work on it, as well as to remove the wheels and move them around (with wheel weights, we're talking a couple hundred pounds - around 90kg - per wheel). The good news is that we have finally finished the shop crane (more on that another time). In the interim we needed a newer lawn mower, so while we were at it we decided to buy a garden tractor that can also push some snow around. Being the "big operators" that we are, we decided on something with some real power, so we purchased a used John Deere 430 (diesel!) with low hours on it. My son Zeke did a lot of work refreshing the machine:
- Changed the oil
- Fixed a problem with "creep" in neutral due to a wore-out linkage
- Sharpened the blades on the mower
- Replaced the glow plugs
- The battery tested marginal so it got swapped out for a new Interstate
- Cleaned the grill (years of grass were accumulated)
- Changed the coolant
- Installed a weather-proof cab with working lights and a wiper blade
- Added a home-made weight attachment to the back
- Changed the transmission oil and transmission oil filter
- Fixed a hydraulic coupler leak
Here's a picture of Zeke out in the driveway getting ready to practice on some sleet we had earlier this month (December 2016). Notice the hearing protection - the diesel is very powerful and very noisy! It's almost industrial equipment.
I also found a great deal on a used tracked two-stage Cub Cadet snowblower and it is sitting in our garage, cleaned up, lubed, and ready for action! Here's what Zeke did on this one:
- Cleaned the carburetor
- Replaced a fuel line
- Cleaned and gapped the spark plug
- Added a wind screen
When we get ready for winter we don't kid around!
As you can see from our example, a few tweeks to technology and you're really in business. I think this applies to 3D NAND. Now NAND flash is nothing new, but the arrival of better yields of 3D NAND allow two very important things: the price comes down, and the density gets better. Not that long ago the idea that NAND flash would ever supplant disk drives would have been ridiculous, as the price differential was simply too high and the capacities too low. But take a look at one of the options for the IBM SVC - a whopping 15.36 TB flash drive in a drive form factor. What's really astounding is that the 15.36 TB drive comes in a small form factor! With TLC NAND driving the cost down, and 3D fabrication driving the capacity up, I see disks getting squeezed into a niche for use only in secondary or archival scenarios.
Modified on by Jack of Maryland
IBM closed on its acquisition of the product and technology divisions of The Weather Company (TWC) today, which brings together the cloud data expertise from TWC and the powerful analytics capabilities that IBM has, as well as the ability to deeply analyze the data using Watson. Does this mean that you will soon be seeing Jack Arnold on the Weather Channel doing the morning AMHQ program, or out in the field with Jim Cantore? Ummm - no. IBM didn't buy the broadcasting part of the business, just the web business and the superb IT infrastructure behind it.
Remember when I said I was worried about a winter storm coming? Well like the title says, "it happened". The offical total I got was somewhere around 30 inches (76 cm), as you can see in the map (look for "Frederick" in the "Northwest of the Beltway" image). In the last post I said our International Cub was in restoration mode - and it still is. So we fell back on our much smaller Cub Cadet lawn tractors. We have two snowblowers, and two blades, one of each for each Cub Cadet. The first snowblower, our prime weapon, a two stage on the newer Cub, failed because the guy who sold it to us had stripped the gear and didn't tell us. Grrr. Then we tried pushing the snow, but it was just too deep. Last we hooked up the other snowblower, only a single stage on the poor old tired model 122 (which was made from 1965-1967). Success! After some shoveling to get a path to the top of the driveway (it lost traction on the slope) we were able to run it down the driveway and blow enough snow to open up a path for us to get out. Once the snow started melting it was too heavy to blow anymore so we used the hydrostatic drive on the newer Cub to push a blade around to do mop-up work. Here's a picture of the mighty model 122 earning its spot in history by saving my family from being trapped in the blizzard of 2016.
Modified on by Jack of Maryland
It's certainly an interesting time to be in the storage business. EMC has given up and decided to throw in its lot with Dell. I'm a bit surprised on this one. Even if you value all the storage hardware and software at zero, there's still a lot of interesting assets at EMC, starting with its stake in VMware. I think this might have more to do with Joe Tucci wanting to leave on a final note rather than wait to see what a successor could do. Some folks are being shown the door at EMC's subsidiary, always a sad thing.
IBM might have something to do with that. The better alternative to VCE (and FlexPod for that matter) is the IBM-Cisco VersaStack, which uses IBM V5000, V7000, or V9000 storage in conjunction with Cisco UCS compute and networking to provide a converged solution.
HP takes a different tack - instead of combining, it is splitting up. Although to be fair, the x86 servers and the storage are still together.
Hitachi keeps plugging along, Moshe Yanai has another startup (Infinidat), and there is a million tiny flash vendors (and some of those will be gone by the end of 2016, in my opinion).
IBM has started shipping the mighty DS8886 and the entry DS8884 in December. SVC is still the leading storage virtualization product, and nothing beats XIV for ease of use. We have a number of enhancements in the pipeline for 2016. I can't wait to talk about them!
There's been a little hitch in the restoration project we've embarked on with our 1949 Farmall Cub. Another one of my son's projects (yes, he has multiple at any given time) is a 1968 Ford F100 pickup (which you can see on the back right in the picture below). Well, it had a little problem. The alternator conked out so we had to get a new one, which then didn't fit the mounting bracket. Using his trusty Millermatic® 211 MIG Welder he crafted a new one, and then was able to mount the device properly. All is well now with the
429 428 cubic inch engine (we can't claim credit for that one). Now the solstice has come and gone and its darned cold here in the Maryland region of the Northern Hemispshere, and the shop is not heated, so the restoration might be on hold until spring. The backup plan is to use another project, the trusty Cub Cadet 124 122 in the foreground of the picture, in the interim.
IBM has had a couple of interesting announcements lately, updating some products that we've had for a while, giving them a new look and new capabilities. First, let's talk about the XIV announcement. The XIV was an IBM acquistion back in the beginning of 2008. Since then it has delighted customers with the best storage GUI in the business. We had added the ability to do compression from another acquistion (more on that in a minute) but there were times that the overhead of the compression threads would slow down XIV throughput. The model 314 largely eliminates that problem by doubling the number of cores (going from 1 x 6-core CPU to 2 x 6-core CPUs per module) and doubling the amount of RAM from 48GB to 96GB per module. Remember that the XIV can have up to 15 modules and you quickly see the tremendous boost in compute resources that the 314 is giving us. The original XIV came with 6-15 modules, where this one is restricted to 9-15. The SSD size is limited to 800GB per module (there used to be an option for 400 or 800GB).
So where did the compression software come from? Now we have to turn to the next product which has been updated. IBM acquired Diligent in 2Q2008 (a busy year for IBM storage!). That technology was placed into our TS7650G deduplication gateway which we have sold for some years now. Well, now comes another recent announcement of a TS7650G upgrade. The XIV model 314 is already shipping, but the TS750G changes mentioned are scheduled for 2016. There is going to be a huge space reduction for the base unit from 7U to 2U, and even better the clustered version goes from 14U to 2U! The current limit of backend storage used by the TS7650G as a repository will be doubled from the current 1PB to a whopping 2PB. IBM has also stated that the product will be available in a software only version (continuing to enhance our industry-leading portfolio of software defined storage).
Speaking of making old things new again, my son has been diligently working on our 1940s Farmall Cub. So far he has accomplished the following:
- rebuilt the carburetor
- rebuilt the generator
- removed rust from the gas tank, sanding it, and putting Bondo on to fill dents
- repaired a radiator that was cracked at the top, via soldering, adding JB Weld, scuffing it, then painting black
Here's a 'before' picture of the project and my son in his shop. Notice that the tractor has a nice patina of 50+ years of rust and enough accumulated grease and oil secretions to make us a major player in OPEC.
End of An Era
One of the true giants of the computer industry, Gene Amdahl, passed away this week. He was a hero of mine for his technical brilliance and entreprenuerial spirit. Although he was a competitor for a while, I always admired him and wished him the best (except for whenever he was selling against IBM, of course). The obituary says that none other than TJ Watson Jr called Gene the "father" of the S/360 - a huge milestone in computer development.
The z13 is a direct descendant of the machines that Gene developed back in the 60s.
Speaking of an end of an era, I'd like to note another key development from IBM. For the first time ever IBM came out with a top of the line storage system in a 19 inch (482.6 mm) rack. Back in the days of the 3330, 3350, 3370, 3380, and 3390s the drives looked like a washing machine. As IBM storage transitioned to use RAID concepts and smaller drives we came out with the ESS E, F, and 800 models starting in 1999. These were commercial refrigerator ("double wide") size. The successor was the DS8000, launching in 2005. This was still a pretty good-sized machine, about the size of a big residential refrigerator. In October we announced the DS8880 series, which is rack mounted (by the way, this ships in an IBM rack - if you want it in your rack and plan to order a few dozen come talk to us. Talk to us in any case - we're always listening.). Now I know that we had the late, lamented DS6000 which went end of service in September of this year, but that was always a midrange device, not meant to be the king of the hill.
In my last post I talked about something to buy to keep to clear the snow off my driveway this winter. With the solstice only 5 weeks and a couple of days away it's time to start taking the matter seriously. So my son and I went out and bought a 1949 Farmall Cub with a front plow. This is definitely a restoration project. We think we are the third owners of this venerable iron. My son is clearing rust off, rewiring, cleaning the engine, you name it. I'll put up a picture or two of the project soon.
I have a tough decision coming up and need better weather forecasting to know the right answer. The Washington Post "Weather Gang", who I really like for their well thought out predictions, just came out with the following headline: "History shows a strong El Niño may mean huge snow for D.C. this winter or almost none". Wow, that really clears it up!
I could ask IBM's Deep Thunder what the forecast is for this afternoon, but it's not designed for long range predictions. So I'm on my own - do I buy the snowblower or small tractor with a plow - or just hope for the best? I'd hate to get a nasty surprise come the middle of January.
We all got a very pleasant surprise from IBM this week. The IBM mainframe has been around since 1964 (and thanks to TJ Watson for "literally betting the company" on the development of that gem). The big surprise is that for the first time IBM came out with a top of the line mainfrrame that is not running MVS as the primary operating system (yes, I know they call it z/OS these days, but isn't it really MVS release 20.something?). We've embraced Linux whole hog with our IBM LinuxONE™ platform. Now the customer has a tough choice - do I run Linux in LPARs, on z/VM, or KVM? My opinion - run it on z/VM, for the powerful tools it gives you to work with Linux guests. Now do I run with FB or ECKD DASD? As my colleague Richard Lewis so eloquently puts it, "In the case of z/VM all ECKD DASD devices that have been made available to the LPAR via the IOCP are automatically detected and available for use. Any operating system including the control program of z/VM does not need to do anything except start an I/O channel program in order to use the device. There is no need for path management or anything like that as the z System I/O subsystem takes care of all that." Richard did a great demo of that for one of my customers - see your local IBM rep to get Richard to show it to you as well.
OK, so I don't know what to buy, if anything, to keep my driveway clear during the coming winter, but as for folks who want to run a lot of Linux guests very efficiently - it's IBM LinuxONE with z/VM.
I have a pet peeve I'm going to talk about today.
There are a lot of cases of mis-used and over-used words, and one that really drives me over the edge is the term "superstar".
For one example, let's look at a news item on a country singer's tour bus getting getting destroyed in a fire. The article refers to the singer as a "superstar". Wait a minute - talented, yes, perhaps fairly well known in country music circles, but a superstar?
Let me give you my definition of a superstar - it's someone who is so famous that people who do not follow whatever area of endeavor the person is in still knows them. So for instance, people who don't follow golf still know who Tiger Woods is. Folks who wouldn't know a basketball from a water polo ball still know who Michael Jordan is.
Another I would put in this category is Frank Sinatra. People don't have to know all the details of his singing and acting career to know that he was really something special.
Are there examples in the computer field of machines that hold this sort of fame? I would argue yes - examples being the IBM System 360 (the original "mainframe"), the IBM 3420 tape drive (which still shows up in movies from time to time as a "computer system"), and of course the beloved IBM PC - right down to the rose.
How about superstar systems for folks who do know what a computer is? I would argue that us techies would vote for the IBM POWER chip (which swept all other *IX systems from the marketplace), the ESS/DS8000 line (which revolutionized speed and reliability of large storage systems), and so as not to be parochial, the DEC Alpha chip and the CRAY-1 supercomputer, both astonishing technical advances for their day.
The humidity is dropping to humane levels where I live in Maryland in the United States, so I may get to see some stars tonight myself - Polaris, Sirius - real stars.
Let me put in a quick plug for IBM Systems Technical University 2015, which you can sign up for here. The event itself is from 05-09 October 2015 and coming up fast, so if you're not already registered, please do so. You won't regret it!
The picture is from an ornamental tree where I work.
It's been a tough winter here in the Northern Hemisphere. Just a few weeks ago I survived getting pummelled by a powerhouse storm with about 9 inches (about 23 cm) of snow. The snow never came down fast, but it just kept coming, hour after hour. Since it was winter, I expected that precipitation would come down as a solid, but the depth of the snow was a bit of a surprise. We were first forecast to get 2-4 inches (5-10 cm) but as the storm got closer it kept intensifying until we wound up with a big pile of it.
Time has moved on and now the delightful season of Spring is here.
Not only are flowers and buds appearing everywhere, but IBM announcements are coming quickly as well (I was almost going to say that they were coming "Fast and Furious" but that would imply a tie-in to the movie which I felt would not be right since I haven't seen it yet).
I've been asked by one of my customers to comment on the recent Spectrum Storage announcement. Let me lead off by saying that I'm really not a fan of renaming products. For a while there renaming IBM storage products was all the rage. The 3592 became the TS1120. The FAStT900 name was retired in favor of the DS4500. The venerable 3584 morphed into the TS3500. I have to tell you that confusion reigned in the hallways of IBMers, business partners, and customers alike. When I worked in the Tucson Briefing Center customers would come in and tell me they wanted to hear about the newest features of, say, the 3584. Then they would look on the agenda and see "TS3500". That always lead to an interesting conversation where I would have to explain that while, yes, they did have a 3584, that talking about what appeared to be a completely different product was in fact what they has asked for.
Despite my aversion to calling the same thing by two names, I have to admit that I like the Spectrum moniker. As wikipedia says it, a spectrum is "a condition that is not limited to a specific set of values but can vary infinitely within a continuum". It's not infinite, but we do have a lot of great products:
- Spectrum Scale (GPFS)
- Spectrum Virtualization (SVC)
- Spectrum Accelerate (XIV software)
- Spectrum Control (VSC without SVC)
- Spectrum Archive (LTFS)
- Spectrum Protect (TSM)
The noted blogger Martin Glassborow dings us for the rebranding, but even he goes on to say "The products they have are generally good...Products that really work!" I have to agree - these products are industry-leading (and in the case of LTFS - oops, I mean "Spectrum Archive", good enough to win an Emmy). So while I agree with him that renaming products might make for some confusion in the short term, in the long run it was the Bard himself who said "that which we call a rose..."
Let me put in a quick plug for Edge 2015, which you can see here. The event itself is from 11-15 May 2014 and coming up fast, so if you're not already registered, please do so. You won't regret it!
Modified on by Jack of Maryland
The plants are changing fast outside now. Here's a picture of my little guy showing how big the forsythia is up in our cul-de-sac. This is one of my favorite flowering trees (and of course he's one of my favorite kids too). Anyway, trees and flowers are blooming left and right here.
Speaking of things moving fast, I had the pleasure of working with Paul from Aspera this week. This is an extremely useful program that moves data fast - and I mean REAL fast, up to 100x traditional methods. Normally data is moved across TCP connections, which have served us pretty well for decades. The problem with TCP is that it is too quick to scale back if a packet is dropped, and too slow to scale back up once it deems the network is OK. Enter Aspera. The patented protocol used in place of TCP is not only fast, but it is reliable, and secure (yep, it can employ encryption should the customer desire).
You can read all about it at the Aspera website. And how good is it? It's so good that IBM bought the company. If you need more information, please feel free to contact me and I'll put you in touch with some of the folks working there.
And let me put in a quick plug for Edge 2014, which you can see here. The event itself is from 19-23 May 2014 and coming up fast, so if you're not already registered, please do so. You won't regret it!
Modified on by Jack of Maryland
Spring has finally sprung in this part of the Northern Hemisphere and the plants are blooming right and left. In the past few days I've seen daffodils, crocus (croci?), and snow drops. The birds are out and singing - I've seen a very pretty cardinal flying around our house. And let me put in a quick plug for the Early Bird discount at Edge 2014. The event itself is from 19-23 May 2014, but the Early Bird discount ends on 20 April 2014, so don't delay.
Even so, we're having a bit of a tussle between the climate and the weather. The climate is the long term trend - normally it will be around 70F (21C) this time of year. However, it was much warmer over the weekend and today it is much colder. Climate is a gradual process, while weather is what is happening outside NOW.
Turning to technical matters for a moment, I want to explain the difference betweek Easy Tier and I/O Priority Manager (IOPM), both of which are available on our flagship DS8870. Both of them modify the performance of applications. Easy Tier aims to migrate data to the appropriate tier of storage. IOPM is designed to give maintain the best possible response for key applications. They sound similar, but here's the difference. Easy Tier is, by design, a gradual process. We move up to 8GB every five minutes, so as not to impact the customer applications. It's meant to be a background process, invisible to the customer right away, but gradually making the response time better and better. IOPM is designed to work right NOW, nearly instantaneously giving a boost to the most important applications. The customer defines which LUNs or volumes should have the best response times. IOPM constantly monitors I/Os to those LUNs and volumes, and should they be falling short of where they should be, then IOPM will delay lower priority I/Os to give more system resources to the key applications.
For more on IOPM, read this redbook. For more on Easy Tier, see this redbook (and note that a draft version was available as of this posting).
Happy spring to all! (or fall depending on which half of the world you live in).
Modified on by Jack of Maryland
We recently rented Thor: The Dark World and spent a long evening watching it. I don't mean to sound like it was particularly bad; it's just that the first movie was really visually stunning, with a decent story, while the second - well, it just had neither. It felt like the CGI team skimped a little bit on the depiction of Asgard in this one. Heimdall turns into an ordinary character - yes, his role is bigger, but not better. The budding love interest of Sif never takes off. I found that part of the story very disappointing, as it just not believable. I could go on, but I don't mean to sound so negative. It was actually a good night's entertainment. It's just that I expected more.
It's actually more fun talking about the disruptive technology that enables me to watch Thor at home. The Redbox company blew up the previous model of going to a neighborhood store such as Blockbuster. Now you can pick up a DVD at a huge array of convenience stores, drug stores, and even at the king of commerce, Wal-Mart.
Speaking of expecting more from something, IBM notes that infrastructure matters. We're only two months away from the storage event of the year, Edge 2014, and I want to give a heads up on a really interesting topic that will be presented by my colleague Paul Schena. The subject is still confidential at this point, since it hasn't been announced yet, but it will involve some really great technology in our SONAS system. SONAS has at least two great capabilities that sets it apart from other file servers: first, the ability to easily store and serve millions of files, and second - well, that one you are going to have to wait until you hear Paul's presentation. Whose infrastructure you choose does matter, and Paul will show at Edge a capability that no other vendor can match. Be there or be square!
Continuing on the theme of Frozen from last time, it's still freezing (literally) here in the state of Maryland, USA. The temperature today is supposed to crawl up to 32F/0C and we'll be lucky to get that high despite the abundant sunshine. At least the trend for solar gain is headed in the right direction. We're now exactly one week away from changing over to the half of the year when the sun is strongest in the Norther Hemisphere, thanks to the Vernal Equinox.
And speaking of changing over, I had an interesting exchange with my buddy Byron from our ATS group (and I'm putting a plug in for their great set of webinars). The question came to me about changing a disk system over from direct connection (via a SAN typically) to servers, to being virtualized behind the IBM SAN Volume Controller (or other Storwize family members, such as the Storwize V7000). Does the SVC require that we have to reboot the server? The answer is no, the SVC/Storwize code does not require that. However, the operating system on the server may need it - say if it is changing multipath drivers, then it would be necessary. So the answer is, our code doesn't require it, but check to see if the server can handle it.
As for me, I could use a change in the weather. As my brother Fred says, "Stupid Groundhog!".
I have another movie recommendation - but it's not one that I've seen (yet). My family went to see the new Disney movie Frozen and came back thoroughly enchanted. To me that means more than a dozen reviews from stuffy critics. In terms of audience demographics, my family ranges in ages from 6 to, uh, I think late 20s for my wife (or very close to that).
Now the title of this story ,and much of the setting, has really hit home for me this winter. We've had massive amounts of snow, and went through the tribulation of an ice storm (knocking power out for about a day and a half). Even a few hours without electricity will remind you of the importance of infrastructure. Here's a picture of my son Zeke clearing our infrastructure - I mean driveway, just this week.
Speaking of the importance of infrastructure, I've been thinking a lot about the data transfer infrastructure for large disk systems (and by disk, I mean both traditional spinning disks and solid state devices (regardless of form factor). John Elliot, who is the lead hardware designer on the DS8000, came to Maryland to speak to some customers last month. He was asked if we plan to support infiniband on the DS8000. The answer is no - not because it's a bad interconnect - we use it internally in our XIV systems for one thing, and in the coupling links for our zEC12 mainframes - but there just hasn't been any market demand for it in the market space that the DS8000 serves.
So what is the dominant interconnect in the storage space? No surprises here - it's still fibre channel. I was asked in the last week what my outlook is for fibre channel connectivity, and should a customer switch to FCoE. My opinion is that fibre channel will continue to dominate through the end of this decade. Today the majority of customers are on 8Gb (yes, I know 16Gb has been available for a couple of years now, but the adoption across the whole datacenter is still underway). 16Gb has some a great advantage in efficiency (64b/66b encoding is 98% efficient vs the 80% of 8b/10b used in 2/4/8 Gb fibre channel), but it's time is not come yet, and won't for another year or two.
So what's the deal with FCoE? IBM has a number of storage systems that support it (see the redbook, "Storage and Network Convergence Using FCoE and iSCSI"). I'm not saying this is a bad protocol either - an interesting study published as a redpaper shows that in many cases 10Gb iSCSI can have equal or even better performance than 8Gb fibre channel.
The point is, I see no reason to rip out a perfectly good fibre channel implementation in order to put in ethernet, whether iSCSI or FCoE. Why - because a vendor says "it's cool"? I think you have better things to do with your money. People running datacenters have built up a huge knowledge base in running fibre channel infrastructures. Those infrastructures are fast and reliable - so why change?
Oh, and about that DS8870 with an infiniband connection: my guess is we'll be glad to build it. Just show me the order for DS8870s, quantity 50 or higher, and we'll get right on it.
My older son asked me to watch "The Hunt for Red October", starring Sean Connery, with him earlier this week, so I did. Now I realize this is not exactly breaking news, seeing as how the movie was originally released in 1990, but I thought it was quite good and I give it my seal of approval as an excellent action movie to watch. I did not see the movie when it first came out - at the time I was busy in Endicott NY working in the group that was nuturing a set of new products, one of which later became ADSM, and was eventually renamed as TSM (Tivoli Storage Manager). Audience participation time: if you can tell me the name it had prior to ADSM I will really be impressed!
Anyway, the point is that sometimes things that started years back turn out to be still quite powerful. 30-some years on TSM is still the best backup product in the marketplace, 26 years after Goldfinger (in my opinion, the best Bond film of all time) Sean Connery was still a great actor, and after almost 10 years (it debuted in 2005) DS8000 is still top dog in the enterprise disk storage space. Here is a link to a good infographic (masterfully created by my colleague Allen Marin) on the DS8000. Please enjoy!
One more little aside on the movie - I was both amused and bemused by the Russian sub commander with a Scottish burr.
Modified on by Jack of Maryland
Back in December I gave a glowing review of the movie The Hobbit Part 2 (technically "The Hobbit: The Desolation of Smaug"). A quick recap - go see it, rent it, buy it - it was very good.
Now for a completely different take on another high profile movie, let me take a few words to describe The Lone Ranger, which we rented to view at home. It was dreadful, awful, a complete waste of money and time, and just downright bad. I'd say more about it but I don't want to take the time to look through a thesaurus since I'm sure you've already gotten the point. The length of this movie is 2 hours and 29 minutes, which is just about 2 hours and 29 minutes too long. I'm struggling to come up with a scene in the movie that I liked, but can't quite do it. I found the movie confusing, boring, and in some ways disgusting. I've seen speculation that people don't want to watch Westerns any more. No, I conclude that people don't want to watch bad Westerns any more.
The reason I got the movie in the first place (yes, it was my idea, though I hate to admit it), was that I fell for the myth that any movie with Johnny Depp would be a good movie. Maybe that was true once upon a time, but I now consider that to be a myth in the bad sense of the word.
Turning to work-related things, there is another myth I've heard recently. Some people have the mistaken notion that the DS8870 (latest version of the fabulously successful DS8000 line) is for mainframes only. Not so! I've seen the figures from our call home data. They indicate that although yes, while the DS8000 does attach to a lot of mainframes, it attaches to many more servers of other types. In fact, the ratio of mainframe attach for DS8000 is roughly 1 in 3. So what is the attraction for non-mainframe folks? Oh, just great reliability, fantastic performance, and really powerful features such as Easy Tier (and Easy Tier is no charge, by the way).
And by the way, if you have an IBM storage box that isn't calling home, please enable it to do so. The benefits are many:
Avoid unplanned outages
Securely and automatically send diagnostic data to IBM Support
Get faster support from IBM
Save time and money maintaining your IT environment
There is no additional charge for systems under warranty or maintenance agreements
See here for more details and a short video.
I want to remind my readers about Edge2014 - it starts 19 May 2014 at the Venetian in Las Vegas, NV. It's not that far away now!
Updated 20 FEB 2014 - added "while" for readability.
Here it is 2014 already and IBM is making some big announcements to start the year off. One of them is in my area of speciality, which is storage. We are (re)announcing the IBM FlashSystem™ 840. This was actually announced back in December, but who had to time to read IBM announcement letters during the busy holiday season? So we're taking the time now to let folks know that there is another great solution in the IBM FlashSystem (formerly Texas Memory Systems RamSan) family.
Let me give you a few highlights of this device:
- 2U form factor- minimal footprint for best of breed ROI
- Low power draw - 625 watts
- Field upgradeable, granular capacity: 4, 8, 12, 16, 20, 24, 32, 40, 48 TB
- Fully redundant and hot swappable architecture:
- Concurrent code load, highly serviceable design
- Low Latency 135us read /90 µs write
- 1.1M IOPS
- 8 GB/s Bandwidth
- 16Gb/8Gb Fibre Channel or
- 40Gb QDR InfiniBand or
- 10Gb FCoE
Another quick look at the product and a link to a draft redbook can be found
Now some interesting things are going on in the storage class memory
space. I still believe in one of my predictions from Edge2013 - nothing is going to take the place of NAND flash in 2014. Why not - there's just too much money invested in NAND flash and continuing to be invested, for some me-too technology to come along and supplant it. I do think it will eventually be shoved aside by some better technology, but not before the end of this decade. By way of illustration, see Micron's withdraw
from the PCM market. I'm sure that PCM is a great technology - IBM has put a lot of research
into it. The problem is cost - nobody has been able to develop a better (cheaper AND faster AND more reliable) technology than NAND flash and I don't think anyone will - in the short term.
Speaking of Edge, this year's event is coming up fast - it starts 19 May 2014 at the Venetian in Las Vegas, NV. Enroll now! Last year's event was sold out, so don't take a chance on missing the premier storage event of the year.
Whew! I got lucky - my kids were so excited that they stayed up late on Christmas Eve, and thus were tired enough to sleep in until after 0800 Christmas Day. Many toys and other useful objects later, the family is back to more or less normal until school starts again. Next up is a blow-out New Year's Eve party (for the family - we'll probably get some snacks out and watch a family-friendly movie).
Since the end of the year is rapidly approaching, I'd like to take the opportunity to briefly talk about one of the year's big stories in storage, flash memory. At Edge2013, which I was fortunate enough to attend, Ambuj Goyal (General Manager of IBM System Storage and Networking), spoke about flash memory "creating a tipping point in the industry" (you can pick up this thread at the 22:18 mark in the video).
I've been championing the use of flash (aka solid state storage) for some time now. I could tell when it was first being made in a drive format factor, way back when STEC ruled the world of enterprise SSDs, that this was going to have a tremendous impact on the storage industry. I knew then that the spinning disk's days were numbered, and it's starting to happen pretty fast. Now I don't think mechanical drives are going away any time soon, but their use will be relegated to bulk storage for lower tiers of data.
I'd like to make a few comments on a story in TechTarget, which is titled "Top five solid-state flash storage news stories of 2013".
They are, with my comments:
- The 2013 flash-storage buying spree was in full force in 2013
This talks about storage companies buying flash vendors to be able to get into the market. As the article notes, IBM was perhaps the first major vendor to do this when we purchased Texas Memory Systems back in August of 2012.
- Startups rule the all-flash and hybrid storage markets
I have to disagree on this one. Sure, there are a lot of vendors in this space now, but who will be left in a year or two? The DS8870, XIV, and Storwize line of products have enormous advantages in terms of server connectivity (mainframe anyone?), replication capability (such as three sites from a single volume in the DS8870), hypervisor integration and so forth.
- The traditional powerhouse storage vendors have been slow to enter the market
I don't think so. EMC was first to market SSDs in enterprise systems, while IBM was close behind with a superior implementation of storage tiering (Easy Tier).
- Solid-state disks (SSDs) are so 10 minutes ago. The future of flash storage is already here
Yes and no. Granted that flash need not take the form of old-fashion disk drives - see the IBM Announcement letter 112-106 dated 04 June 2012 which talks about a "A new high-density flash storage module for selected IBM disk systems, including the IBM System Storage DS8000". But traditional disk systems aren't going away - they are just going to change in form factor, but not in function.
- Consumer-grade MLC flash overtakes SLC flash as the de facto enterprise standard
Old news. IBM storage has been running eMLC for some time now. I think the real interesting news is that a case can be made for TLC flash. Facebook is considering using TLC flash for cold data. I personally would like to see it used in PCs for operating system and applications, which have a high read-to-write ratio. One could imagine a hybrid eMLC-TLC flash much like hybrid drives use flash for speed and platters for capacity. I envision a hybrid flash that uses eMLC for writes (updates to an operating system, for example) and TLC for capacity.
By the way, don't forget that Pulse is coming up soon - 23-26 February 2014 in Las Vegas, Nevada, USA. I've been asked to speak, though I don't think I will actually be able to make this one. If that changes I will let you know.
The excitement is mounting on this end! My family is ready for Christmas to start. I do hope the kids sleep in a bit and allow us to start the celebration closer to 0800 rather than 0500 tomorrow.
Back on the work front, I'm still on duty today. Somebody has to pay for the gifts tomorrow, and I'm not counting on Santa to do it.
In my last entry to this blog I talked a bit about availability being an important attribute of an enterprise array, following up on the question posed by Robin Harris.
I'm going to cover a couple of techniques used by our DS8870 and XIV storage arrays. The DS8870 uses a fairly clever technique called "Smart Drive Rebuild" or also called simply "Smart Rebuild". The DS8000 RAS microcode examines each spinning disk in the DS8870 twice a day. In a RAID 5 array, if the number of errors exceeds a threshold, the data from the offending disk is copied to a spare disk, and then the source disk is retired. The advantage here is that two precious resources are conserved: (1) I/Os on the other six or seven drives in the array that would be required to determine the parity information so the failing drive could be rebuilt, (2) cycles in the POWER processors that would be used to do the reverse parity computation. The really good news is that the more you need it, the better it works. I say that because the technique gives results that are much better than a normal RAID rebuild when the system is under a heavy load and you need every I/O and CPU cycle that you can get. You can see a graph to this effect on Jim Kelly's blog. So what happens if the drive fails during this process? Simple - the DS8870 reverts to a normal RAID 5 rebuild.
Now you're thinking that this can't have any applicability to the XIV - it's doesn't have the normal RAID 5 scheme in its repertoire. True - but it also collects data on disk errors, much like the DS8870, and makes a similar determination on when it thinks a disk is going to fail in the near future. As you may know, the XIV always has two copies of the data scattered throughout the system, in 1MB chunks. When it detects that a drive has gone over the threshold, it makes a THIRD copy of the data scattered throughout the system. Upon the completion of this process the source drive is "administratively failed" (taken offline). The net effect here is that there is even more redundancy when a drive exhibits bad behavior. And we wouldn't want anything bad on our record when Santa is watching, would we?
Modified on by Jack of Maryland
Yesterday, my son (the tall one - roughly 6'4" or 193cm) and I went to see the new Hobbit (or as we say in our household, in light of the author's nationality, the 'obbit) movie yesterday. It was great - much better than the first installment of the trilogy. We paid extra to see it in 3D at an IMAX theater, and we both thought it was money well spent. The CGI was better, the plot was faster paced and more interesting, and the character development more realistic. In particular I was struck by how the dwarves were no longer portayed as buffoons, but rather as beings with dignity and worth equal to humans. My opinion is that to be faithful to the original Tolkien, the dwarves should not be portrayed in the same manner as Disney's Seven_Dwarfs.
To take the analogy from Middle Earth now to my work world, the DS8000 series of storage controllers keeps getting better as well. One of the selling points of the DS8000 line is its reliability. The DS8100/DS8300 was followed by the DS8700 and then the DS8800 and each of them learned from their predecessor and took an already high availability figure and drove it even higher. I get to see some of the field data for the DS8000 line, and I'm happy to say that although all of the models have great numbers, the latest (the DS8870) has the best. In fact, it is almost literally off the chart it's so good. Just what sort of numbers are we talking about? As the IBM DS8870 data sheet puts it, "six-nines availability that has made the integrated System z HyperSwap solution the gold standard for enterprise high availability".
Allow me to throw in a little technical discussion on the "System z HyperSwap solution". It typically refers to GDPS/DS8000 integration (basic HyperSwap is built-in to z/OS but let's focus on GDPS in this discussion). GDPS is a family of high availability and disaster recovery solutions. I'm going to explain System z HyperSwap. The setup here is an IBM System z mainframe connected to two DS8000s. Those two DS8000s are in a Metro Mirror (IBM-speak for synchronous replication) relationship. The mainframe writes to only one of the DS8000s, and those writes are mirrored to the second DS8000 before the write is committed to the server. Should the source DS8000 go down, GDPS quickly switches reads and writes to the second DS8000 with minimal to no disruption to the mainframe workload. How fast? For most clients it is 10 seconds or less - so fast that typically the mainframe workload just pauses, then continues without any disruption. What's the integration with the DS8000? GDPS is an open product - we try to work with everyone. Specific parts of the architecture are licensed to some of our competitors in the storage space (EMC and HDS, for example). When GDPS notices a pause in data flow to or from the DS8000, it has to make a quick decision whether to switch over or wait until the DS8000 recovers. Part of the integration here is that the DS8000 helps with that decision - it provides the GDPS code with information on how much longer the DS8000 thinks it will take to recover. This is a DS8000 exclusive and is one more reason why our DS8000 is the best choice to work with IBM System z mainframes.
Back in November, blogger Robin Harris posed the question of what workloads were appropriate for enterprise arrays. "Availability" was judged to be the best answer (though I would argue not the only one), and this is one area where the DS8870 stands head and shoulders above the rest.
I had very interesting and productive week, and learned a few things in the process (I always do). On Monday I went to a customer location to upgrade two V7000 controllers from a 6.4 level up to a 188.8.131.52 level. It's an amazingly simple process, with a few minor quirks. It would have been very easy if I hadn't made a little mistake along the way.
The first step in the process is to download two packages - one to check that the upgrade is able to proceed, and then the actual upgrade itself. In this case, the upgrade test was "IBM2076_INSTALL_upgradetest_11.1" (2076 being the machine type for IBM Storwize V7000) at a size of 52k, and then the big one, "IBM2076_INSTALL_184.108.40.206" at a size of 326.5MB.
I hooked into the customer network, brought up a browser (Firefox is the preferred choice here) and initiated sessions with the two V7000s. All went well with the first upgrade - about 20 minutes for the first canister, a pause, and then the second canister upgrades as well. I kicked off the second upgrade, then made a little mistake as to which screen I was on (the two IP addresses of the V7000s ended in .20 and .120, so you can see where it could get a little confusing). It appeared to me that the machine was hung, when in fact it was proceeding OK. I rebooted it, which stopped the upgrade process. When it came back online it said "The upgrade has stalled, click to revert to the previous version of the software CMMVC6054E", I tried to take it back to the prior level, but as it was in the middle of restarting the upgrade process it refused to do so. I'm happy to report that despite my efforts to confuse it, the V7000 did revert back after a short time. I then very carefully started the upgrade process again and it ran to completion. Near the end it was a little unsettling to see the main screen display the old level of microcode while the upgrade screen said the system was at the new level. However, less than a minute later the two synchronized to the same level (220.127.116.11)
I learned two things - when working with the Storwize code, be patient! Sometimes it takes a while to process. Second, this is a very resilient system. It had some great checks built into the system to understand where it was in the upgrade process and to not let a knucklehead user (me) derail it.
We (the customer and I) did find a few usability items that I thought could be improved. First, when displaying the status of long-running tasks there are two columns. The first contains a number, and the second the name of the task. That first unlabeled number was the minutes left to completion - I'd recommend a column heading to that effect. Second, when trying to use all the storage in a managed disk, we had to interpolate to find the number of megabytes that would completely use it. A "use all available space" button is needed here. Last, we'd recommend that the upgrade process itself be put into the tasklist so that we could see the status from that view.
In summary: the system is bullet-proof. My hat is off to the developers for creating such a resilient storage controller. I'd recommend a few minor usability changes, but the system runs well and is easy to manage.