|
|
| |
 IBM mainframe makes it possible
| | |
 |
|
Overview
Hewitt Associates depends on System z servers to flexibly and securely handle its large, complex — and growing — workloads. |
|
 |
|
Challenges
Hewitt Associates right now provides on-line services for administering 18 million employees of our clients worldwide. We're bringing online new services for existing clients and that takes an ever-growing amount of infrastructure to support. So this fall, when we host our annual enrollment facilities for many employers, we'll expect over 13,000 concurrent employees on those sites at a given time. |
|
 |
|
Resiliency & availability
We use System z as the backend database and the business services engine to support all of it. We're able to make the changes we need to in the infrastructure as we go, while we continue to do service. We don't bring things down in order to maintain and upgrade facilities. |
|
|
|
Security
The System z has one of the most advanced models of security of any of the computing platforms available today. We've been relying, in the last year or so, on the crypto facility inside of System z, so that participant access is guaranteed to be only done by that participant. |
|
|
|
Simplified infrastructure management
Especially over the last few years, as we migrated into 64-bit, we're able to support spectacular amounts of work on a very simple to administer, straightforward, clear basis. So that we have a very predictable set of outcomes. We know how we're going to handle new pieces of business and new workloads and we don't have to rework the infrastructure. |
|
 |
|
Systems integration
The basic software stack that we run on our systems that are z/OS, would be DB2, CICS. We run WebSphere in distributed running into System z. We use MQSeries to interconnect a lot of our distributed and mainframe systems. |
|
|
|
|
Virtualization
To build a test Parallel Sysplex would have normally required lots and lots of computers, lots and lots of cables, lots and lots of interconnections, and an attendant amount of, a tremendous amount of, complexity. Instead of doing that, what we decided to do was put up a logical partition with z/VM and run the entire Parallel Sysplex configuration entirely virtualized. We were able to triple the size of the stress testing environment, run our test and put them all back the way they were before without actually bouncing or stopping any of the work on any of the neighboring machines or inside of the stress testing environment altogether. It was completely done dynamically, which quite amazed our friends in distributed systems as well as delighted our business people. |
|
|
|
|
Systems management flexibility
The whole on demand initiative from IBM has been a very exciting set of opportunities. We've found that On/Off Capacity on Demand was a means that we could just make that resource available for the days that we planned, and that we would have that capacity available on any day where we had surprising loads that we really hadn't anticipated, without having to put in the capital expense of having that available all the time just to handle a small number of days. |
|
|
|
Business integration deployment
We have very scalable, very robust systems that facilitate all of our major business activities through there, and System z is instrumental in coordinating things with Web servers and app servers, with the voice response units, and all the other kinds of systems that do that work, which establish all of the other elements of our fabric. |
|
|
|
Future deployment
Our goal is to remain invisible. The only way you get visibility in the building of infrastructures is if you really make a bad mistake. |
| For large, complex workloads that are going to grow over time, and require high availability, there is no second choice to System z. We've evaluated all alternatives and found that even though one could do alternatives, the decision is that they would just be bad ideas. |
|
|
|
| |
|
|
|
