Title: | DEC Rdb against the World |
Moderator: | HERON::GODFRIND |
Created: | Fri Jun 12 1987 |
Last Modified: | Thu Feb 23 1995 |
Last Successful Update: | Fri Jun 06 1997 |
Number of topics: | 1348 |
Total number of notes: | 5438 |
Hello, I have a customer interested in determining how difficult it would be to migrate (pickup and move with minimal changes to the application) an Adabas/Natural/Complete application running on an IBM 4381 running DOS under CICS to the VAX platform. The customer has minimal information available to him about the details of the application because this will be a power play move on his part to migrate the application from its current home (big blue haven) to the sometimes treacherous waters of VAXdom. So, my customer would like to know the following: 1. Has Digital or its customers ever done this before? If so with what amount of success? 2. The application is using abnout 3 Mips of an 8 Mip 4381, how much VAX will he needed in terms of VUPS? (His question, not mine). Please save your breath on the speech about MIPS, VUPS, comparisons of the two and the importance of throughput as the real measurement of acceptable performance: I understand and agree, HOWEVER, if anyone has information in general about how much of an IBM the application used to use and then how much of the VAX the migrated application now uses we may be able to give the guy a feel for his task. I've already set the stage that these types of comparisons are greatly lacking of any real significance. He understands and does not plan to purchase a processor based on a statement about comparative Vups to Mips, but does need to have this information to convince the rest of the corporation (IBM-oriented folks) that he is getting a much better deal using the VAX. 3. He likes Rdb and has RTO at his site. However, to make his coup attempt successful he is somewhat concerned about the time it would take to convert the Natural application and the database to Rdb. So two questions: a. How quickly could an Adabas/Natural/Complete application be migrated to the VAX using the same products/tools? Any known gotchas? b. How would that same migration proceed if the database were to be Rdb instead of Adabas? 4. They use Complete (Adabas's TP Monitor, I think) on the IBM. What types of problems might this cause for migration of the application from IBM to VAX? If usinng Adabas and Natural or if using Natural and Rdb? 5. Is there anyone who has done some portion of this before? Digital employee or customers are acceptable. Thank you, Rick
T.R | Title | User | Personal Name | Date | Lines |
---|---|---|---|---|---|
1062.1 | Software AG? | TRCOA::MCMULLEN | Ken McMullen | Sat Jan 25 1992 14:46 | 7 |
Your questions are more appropriate for Software AG. They are the people who should have the most experience on both the platforms you mentioned. Also, try contacting Chris Butler in Fredricton, New Brunswick Canada (he is in ELF). A customer of his recently put their Natural/Adabas applications on the VAX platform. | |||||
1062.2 | Details from Software AG | FHOPAS::BREWMN::BREWIS | Tue Jan 28 1992 21:10 | 26 | |
Ken, Thanks for your reply. I contacted Adabas right after entering this note. The contact said that Natural/Adabas applications port very easily from the IBM to VAX environment without modifications to the Natural application or the database. The contact also indicated that the application migration can occur independent of Software AG's Complete product. Side note: Complete is like TSO and/or ROSCOE on the IBM system. IT runs independent of CICS (in other words, not as a CICS transaction). Complete is available on the IBM platform only. My contact said that they use a product called Convert to read the IBM EBCDIC tape to convert the Natural Application and the Adabas database to ASCII. Once Convert is completed, just compile and complete a few manual steps to finish the setup. Most of these center around the creation of directories, subdirectories, logicals, and symbols. Performance Natural/Adabas performance is purportedly comparable to the figures we publish for active (not subscribers) ALL-IN-1 users supported by our processors. He recommended about 1/2 MB memory per user and, of course, the usual cautions about spreading database I/O across as many spindles as you can. Rick |