The Power Systems i 570 Versus Its Predecessors
September 29, 2008 Timothy Prickett Morgan
Throughout the summer, The Four Hundred has been examining how the Power Systems i Edition machines in rack and blade form factors compare to their predecessors and, for entry and midrange machines, compare to Windows boxes. This week, we climb one step higher on the Power Systems ladder and compared the System i5 570 machines from 2006 with the new Power 570 machines of 2008. Basically, IBM is offering more performance for more money for customers who need it, or the same performance for a little less money.
We all know, of course, that last July IBM announced an interim Power6-based System i 570 box, which was a precursor to the revamped product line that came out in April. As I said in the wake of the announcement last July, IBM didn’t hurt itself in terms of delivering price cuts for enterprise customers buying 570-class boxes. What IBM has done, however, is let customers start with a bare-bones machine and add i5/OS or i with or without the database activated (it is called i Application Server when the database is not licensed) and let customers buy 5250 protocol only when they need it, rather than messing around with governors or the Standard Edition (no 5250) or Enterprise Edition (5250 activated on all processor cores) dichotomy. What this means is that for a customer who wants only a few cores activated with the i operating system and then the others for AIX or Linux partitions, a good argument can be made for server consolidation both on a technical and economical basis. This was not the case with earlier generations of 570 boxes.
Having said that, if you compare 2006’s System i5 570 using Power5+ cores and running i5/OS Standard Edition to 2008’s Power 570 i with no Enterprise Enablement (that’s the current name for 5250 processing capacity licensing), or System i5 570 Enterprise Edition machines to Power 570 servers with 5250 capacity on all the cores in the box, you probably won’t be all that impressed with the bang for the buck the new machines offer in certain configurations; in some configs, the improvements if you look at it on a cost per user or cost per transaction basis are pretty good. As the 570 configuration table that I built shows, it really depends on the comparison that you make. I did a bunch between old and new machines to give you a feel for the numbers.
The important thing to remember as you look at these numbers is that even though IBM has aligned 570 hardware prices with the AIX side of its business, it has jacked up the price of i5/OS V5R4 and i 6.1 licenses on each core in the box, too.
The System i5 570 boxes in the comparison are all using the fastest 2.2 GHz Power5+ cores IBM could deliver in 2006. The 3.5 GHz Power6 processors offer roughly equivalent performance to the older processors, so I made a set of 5250-disabled and 5250-enabled configurations based on these 3.5 GHz Power6 chips. These are the slowest Power6 cores IBM offers in the box, and not coincidentally, they are the least expensive processor boards and core activations, too. I also did a second set of comparisons of Power6 570 machines, but using the fastest 4.7 GHz cores. These system boards are a lot more expensive, and processor core activations, at $23,000 per core compared to $9,100 for the 3.5 GHz chips, are a lot more expensive than the relatively modest increase in performance justifies. IBM knows this, of course, and has priced accordingly. You need the performance, you gonna pay.
Unlike the prior Power 520 and Power 550 machines, I have not attached disk arrays to these enterprise servers. I did toss in two disks, a RAID disk controller, and a suitable tape drive, but I didn’t want the price of old and new disk arrays to skew the relative bang for the buck of these two generations of machines. I wanted to show you the merit of base server configurations, with an appropriate amount of cores and memory activated, waiting for storage to be hung on the box. Each configuration includes either i5/OS V5R4 and the Virtualization Engine hypervisor (old gear) or i 6.1 and the PowerVM Standard Edition hypervisor (new machines); the hypervisors allow basic logical partitioning. I also removed Software Maintenance fees from all of the systems, even though IBM requires customers to buy it, because I think customers should have the choice of maintenance services or not (or IBM or otherwise) and because I want to isolate system costs from support costs.
To help make your capacity planning a bit easier, I calculated the cost per transaction per minute for the machines based on IBM’s Commercial Performance Workload (CPW) benchmark and my conversion of that rating to estimated TPC-C online transaction processing performance. I also added user counts to the configurations, even though IBM does not price the machines based on users (just as I did with the Power 550 comparisons over the summer to older System i iron and to entry and midrange Windows boxes). I happen to think IBM should price all i5/OS and i licenses based on a basic (and low) per-server charge plus a modest fee, and adding a user component to the configuration allows per-user comparisons across the System i and Power Systems i lines and with other servers. I am not suggesting for even a second that the number of users shown is the maximum number of concurrent users a machine can host, or even an appropriate configuration. The user numbers are based on the user counts IBM has for the user-priced System i 515 and 525 servers from last year, then scaled up as processor counts go up. If a an entry Power 520 with a single core has 40 users, then an entry Power 570 with four cores should be configured with 160 users; if a large configuration of the Power 520 has 150 users, then a large configuration of a four core Power 570 should have 600 users. It’s just simple math.
In general, as you will see from the table, the price/performance improvement over the past two years with machines that do not have 5250 capacity enabled is quite low. The largest improvement is on the eight-core configuration. And for 5250-enabled gear, the average price/performance improvement is in the range of 35 percent whether it is based on users or transactions per minute derived from CPW ratings. That kind of improvement was the annual rate of bang for the buck betterment for server platforms during the dot-com boom, so I am not all that impressed, to be honest. Argue hard for discounts, and get competitive bids on Windows and Unix iron even if you have no intention of moving off Power Systems i. In this kind of economy, you cannot afford to not be playing hardball.
The other thing you will see from the chart is that the Power 570s with the fastest processors offer better bang for the buck as well as more performance on a transactional basis, but if your user base isn’t growing, then the improvement can be small or nil. Proper capacity planning is real, real important. Stay out of the 570 line if you can stay in the 550, and don’t go into the 550 if the 520 will do. Each step up the rung of the Power Systems ladder means more expensive processor cores and software licensing per core. It doesn’t have to be that way, of course, but that is how IBM does it.
Now that I have the System i5 570-to-Power Systems 570 comparison done, I can do what I have wanted to do for a few weeks: stack these Power6 machines up against similarly powered enterprise servers using Intel‘s new six-core “Dunnington” Xeon processor for four-socket servers. I am also curious how Linux and Unix boxes compare to the whole Power Systems i lineup. Fear not, we’ll get there.