May 3, 2021
Timothy prickett morgan
Since the advent of file servers in the 1980s, the rise of client / server system architectures in the early 1990s and the commercialization of Internet networks in the mid-1990s, AS / 400 stores and those using the progeny of this venerable midrange IBM computer had hybrid computing platforms in the data center. That is, a mixture of processor architectures and operating systems other than OS / 400 or IBM i that were in some way associated with or actually doing critical work.
In fact, as you all know, overall there is more raw computation in X86 or RISC servers adjacent to IBM i platforms than in those IBM i platforms themselves. Sometimes it’s multiples like 2X or 3X or 5X, and sometimes it’s one or two orders of magnitude. Typically, however, IBM i systems run at high utilization for transaction processing, batch processing, or data analysis, while the surrounding layer of machines performs application service, file service. and other workloads, but not running at peak. They have excess capacity and redundancy to make up for some of the systems shortcomings, making them inexpensive and yet underutilized and therefore when you do the math, expensive in another way.
We’re not here to discuss this just yet, nor to try to convince everyone to move applications from external X86 and RISC machines to Power Systems platforms, let alone port them to the operating system. IBM i. We do this kind of thing all the time. But what we want to do is let you know about some of the changes to the processor architectures of the X86 and Arm servers that are currently or in progress and how this may impact the workloads running where in your data center. data.
Another thing we aren’t going to do right now is go over the details of Intel’s new Xeon SP “Ice Lake” server processors, AMD’s new Epyc 7003 “Milan” server processors, or the “Zeus” Neoverse The V1 and “Perseus” Neoverse N2 server designs from Arm Holdings, the company that licenses the intellectual property of Arm CPU. If you would like to see the architectural details of these new processors, the following links from The next platform, where I also work, will be extremely helpful:
Intel has had serious issues with ramping up its 10-nanometer manufacturing processes at its foundries, and the years-long delay has allowed AMD, its decades-long rival in X86 processors, not only to catch up with its processors. Epyc, but to surpass Intel’s Xeon. SP in many ways, such as number of cores, raw throughput, and price / performance ratio. And the Arm collective, now led by Ampere Computing and cloud giant Amazon Web Services, and a number of small players with big aspirations in various parts of the market, will be rolling out some really good processors starting this year and onwards. next. The competition between Intel, AMD and Arm is going to be fierce, and unlike in previous years, it is going to remain fierce, with an annual rate of processor updates, the manufacture of transistors for these chips being pushed to its limits and very intense. price / performance competition.
There has never been a better time to buy an X86 or RISC server, and we will be able to say that every year from now on, until Moore’s Law decreases transistors and chips start to get a lot. hotter and much more expensive to manufacture. That day is coming, somewhere between 2025 and 2030. And it won’t be long before what was once in entire racks of systems will be in water-cooled 3D blocks that look like thermal conduction modules (TCMs). of vintage mainframes.
In view of all this, we offer this advice: “Knowledge is power”, as Francis Bacon said more than four centuries ago. Obtain knowledge and use it as leverage for negotiation.
Start a pilot program to study the Arm servers. The easiest and cheapest way to do this is to play with Graviton2 processor based instances in the AWS cloud. These are reasonably powerful and offer very good value for money compared to X86 servers, as we calculated at The next platform. However, you are stuck with Linux on Arm at this point because Microsoft has only ported Windows Server to Arm chips for its own internal use and does not support Windows Server or the myriad of ancillary system software items that run with it on Arm processors. And that doesn’t seem inclined to do that either. But knowing how the Arm server chips work and how well they perform will be a key part of negotiation.
Start a pilot program to study the movement of X86 workloads to Power. It is important. Make a complete profile of all the applications you have run on your X86 iron, and see which are on Linux and can be moved to logical partitions on Power Systems and which are blocked on Windows Server and cannot be moved to Power Systems because Microsoft does not support Windows Server during power supply. You would think that IBM and Red Hat would have already set up an organized and concerted porting factory to do this, but as far as we know, the old migration factory that Big Blue once had to move workloads. Solaris and HP-UX to AIX has not been revitalized to move Linux or Windows Server workloads to Power Systems partitions running Linux. So you have to do it yourself.
Make sure to pit AMD against Intel when considering buying X86 servers right now. If you’ve used a single server builder in the past decade or more, step out of that model. Ask one OEM to offer systems based on Intel’s Xeon SP processors and another to offer systems based on AMD’s Epyc processors. Try to increase the competitive pressure to really drive the bids for the X86 iron. You cannot do this by going to a supplier. They’ll be launching the SP Xeons first, and if you’re a little worried, they’ll make a small deal on AMD Epycs and call it someday.
Buy a lot of capacity now and get a better price in the long run. We don’t generally advocate buying more capacity than you need at all times, but as Jensen Huang, the co-founder and CEO of Nvidia pointed out, “The more you buy, the more you save.” If you make the upgrade of X86 servers look bigger by getting a lot more capacity up front, you get a bigger break.
Use aggressive X86 pricing and competitive pressure from Arm to lower power system costs. Competition between Power Systems and other platforms is indirect, with the exception of the Linux stack, which makes it difficult to get a good deal. But if you can bring more workloads into the deal, you get the attention of IBM and its competitors, and they’ll fight harder to keep the deal going. And, if you do it right, you get a more consolidated, easier to use, and less expensive platform that also includes cheaper IBM i platforms.
It’s win-win-win. Based on what you learn, you will be in a better position to do any movement you may need.
It’s harder to hear the pulse in the server market
IBM i Store Tips for Buying X86 Servers
Take full measure of power servers
Reduce X86 hegemony in the data center
What is the size of all power supply systems?
The collapse of power systems is not as bad as it seems
The ups and downs of the server cycle
HE begins to feel the impact of the big infection
The IT industry could weather the pandemic storm
The midrange gets pinched a little more
“Power First” as IBM quits X86 servers
IBM’s X86 Exit Strategy: Making the Case for the Good and the Bad
What the System x sale means for IBM i stores
Will Big Blue Deep Six be an X86 Biz Server?