Portals eNewsletters Web Seminars dataWarehouse.com DM Review Magazine
DM Review | Covering Business Intelligence, Integration & Analytics
   Covering Business Intelligence, Integration & Analytics Advanced Search
advertisement

Resource Portals
Business Intelligence
Business Performance Management
Data Integration
Data Quality
Data Warehousing Basics
EAI
EDM
EII
ETL
More Portals...

Advertisement

Information Center
DM Review Home
Conference & Expo
Web Seminars & Archives
Newsletters
Current Magazine Issue
Magazine Archives
Online Columnists
Ask the Experts
Industry News
Search DM Review

General Resources
Bookstore
Industry Events Calendar
Vendor Listings
White Paper Library
Glossary
Software Demo Lab
Monthly Product Guides
Buyer's Guide

General Resources
About Us
Press Releases
Awards
Media Kit
Reprints
Magazine Subscriptions
Editorial Calendar
Contact Us
Customer Service

Notes From the Giga Advisor:
Be the Electron in the Wire: Data Warehousing with DB2/6000 vs. S/390

  Column published in DM Review Magazine
April 2001 Issue
 
  By Lou Agosta

One of most interesting inquiries I received last month from a Giga client was a question about key criteria for deciding between data warehousing on DB2/6000 and S/390. Here are the main points of the reply.

The greatest advantage of DB2 S/390 is in reliability, availability and serviceability. Although single image alternative platforms such as AIX- RS/6000 continue to improve, they are still not as bulletproof as the mainframe. Of course, there is a high availability cluster multiprocessing (HACMP) option (using a form of clustering) for AIX. That will at least partly address the reliability and availability issues. Notice also that the initial cost advantage of AIX is thereby reduced.

Data warehousing systems start out being less than mission-critical and requiring a service-level agreement (SLA) consistent with, for example, 12x6 availability (twelve hours a day, six days a week). However, everything seems to slide in the direction of 24x7; and soon the IT staff has to manage a different, more demanding SLA with infrastructure designed for a less rigorous scenario. Obviously, that spells trouble. Whatever the agreement you reach, be sure to document it thoroughly (in writing) and make sure everyone has a copy.

Do you have the capacity - the CPU cycles, the storage, the processing window - on the mainframe? If you have unused capacity - CPU cycles and available storage or either one - on the mainframe, then it may actually be less expensive to use the mainframe. With the partitioning and workload management provided on S/390 (now IBM's zSeries), dedicating a portion of a S/390 environment to a data warehouse is doable. In case you must perform capacity planning and schedule a mainframe upgrade, that will impact cost and schedule significantly. Other-wise, the new AIX system will have an advantage if it is standalone and can be brought in independently of an existing platform.

What is the available storage technology? If you already have system managed storage (SMS) on the mainframe, the lives of the DBAs and IT staff should be made more simple - with fewer outages due to mismanaging storage allocation - due to policy-base storage management. SMS does not, in itself, prevent running out of space (how could it?); but, it reduces the risk of running out of space due to staff oversight in having manually to monitor hundreds or thousands of data structures. Depending on how many data structures are implied by managing 500GB to 1TB of storage, storage technology becomes the critical path. Policy-managed storage is not available on AIX, although what used to be called AdStar Storage Management (ADSM) and is now a part of the Tivoli storage technology solution simulates some of the features.

If you have 1TB of user data, not including indexes, aggregates, etc., then you could end up needing multi-terabytes of storage and a correspondingly higher price point. If you think the data volume is going to grow significantly beyond 1TB or CPU cycles ramp up quickly in the face of long-running, complex queries, and assuming you wish to remain IBM-centric, you may want to look at IBM SP hardware using DB2 UDB (extended enterprise edition). This is the share-nothing architecture which is required for very large systems at the multiterabyte level. Regardless of the underlying platform, we are talking mainframe pricing here, are we not? (Examples of pricing can be found under TPC-H on the Transaction Processing Council Web site at www.tpc.org.)

What and where are the skills of the DBA staff? If they are familiar with DB2 UDB utilities on AIX, then that is a significant advantage for AIX (or vice versa). Managing backup, recovery and reorganization requires an understanding of data integrity in a variety of forms. DB2 log archiving and managing of those logs for system recoverability is (relatively speaking) a piece of cake on S/390. Supplementary storage technology is required on AIX - see the earlier paragraph about ADSM. DB2 S/390 has supported online reorganization since version 5, so 24x7 availability is feasible from a database scheduling perspective. However, DB2 on AIX/NT does not support online reorganization and will not do so until at least the next release. So, that is another availability advantage for S/390. In other respects, ease of use (usability) is what you already know. There are now good third-party tools available for database monitoring and fast utilities on non-MVS platforms. However, the mainframe is still the most heavily instrumented and instrumentable platform - at a corresponding cost.

Since most mainframes are, in fact, maxed-out at 100 percent utilization running transactional workloads, maybe the performance advantage is indeed with the single image AIX box dedicated to its task. However, that is a function of a given environment, not of the intrinsic capabilities of the processor configurations. In short, clients should prepare to load up on main memory and allocate large volumes of buffers to DB2 (or any database, for that matter) regardless of the underlying platform. Undertake some role-playing and be the data, be the electron, in the wire. I/O is the slowest path through the system - and the fastest I/O is one that never occurs because the data is already cached in the buffer.

...............................................................................

For more information on related topics visit the following related portals...
DW Design, Methodology.

Lou Agosta is the lead industry analyst at Forrester Research, Inc. in data warehousing, data quality and predictive analytics (data mining), and the author of The Essential Guide to Data Warehousing (Prentice Hall PTR, 2000). Please send comments or questions to lagosta@acm.org.

 

 

Solutions Marketplace
Provided by IndustryBrains

Embarcadero ER/Studio 6.6
ER/Studio delivers next-generation data modeling. Multiple, distinct physical models based on a single logical model give you the tools you need to manage complex database environments and critical metadata in an intuitive user interface.

Data Mining: Strategy, Methods & Practice
Learn how experts build and deploy predictive models by attending The Modeling Agency's vendor-neutral courses. Leverage valuable information hidden within your data through predictive analytics. Click through to view upcoming events.

Help Desk Software Co-Winners HelpSTAR and Remedy
Help Desk Technology's HelpSTAR and BMC Remedy have been declared co-winners in Windows IT Pro Readers' Choice Awards for 2004. Discover proven help desk best practices right out of the box.

Dedicated Server Hosting: High Speed, Low Cost
Outsource your web site and application hosting to ServePath, the largest dedicated server specialist on the West Coast. Enjoy better reliability and performance with our screaming-fast network and 99.999% uptime guarantee. Custom built in 24 hours.

Get SAP Technologies Training on DVD
For the first time ever, access SAP Technologies Training at your convenience with the TechEd '04 DVD. Each package includes 100s of hours of SAP training lectures & hands-on workshops.

Click here to advertise in this space


View Full Issue View Full Magazine Issue
E-mail This Column E-Mail This Column
Printer Friendly Version Printer-Friendly Version
Related Content Related Content
Request Reprints Request Reprints
Advertisement
advertisement
Site Map Terms of Use Privacy Policy
SourceMedia (c) 2005 DM Review and SourceMedia, Inc. All rights reserved.
Use, duplication, or sale of this service, or data contained herein, is strictly prohibited.