Mark, What's their definition of "sync"? We (this community) have an assumed definition -- a very good one, the Sysplex definition -- that for a variety of reasons makes much sense. That said, if you look at other, lesser servers they don't share that definition. They don't have Sysplex.
You can still do somewhat better than those other servers. (The clocks are much better at least.) You can have the z10 get its time separately from the 9037s and thence from an ETR, and the z196s from STP thence its ETR. If the ETRs are the "same"(*) then the z10 and the z196s will have "similar" times which should be bounded within some statistical distribution. Is that good enough? It is for lesser servers. Or it isn't good enough even for lesser servers, but that's all they can typically do. That arrangement also means the z10 and z196s wouldn't be in the same Sysplex, with associated financial implications presumably. Maybe time to ring up IBM for a pair of zEC12s and replace 3 with 2, for example? Both z10ECs and z196s are MES upgradeable to zEC12s. (*) It can't be *exactly* the same even if you're asking the same authoritative time source (e.g. NIST) because you're asking via different paths at different times in different requests in different formats. But at least you could ask the same authority (e.g. NIST) to knock out one of the variables, for what it's worth. The 9037s might ask NIST via dial-up modem, for example, while STP would ask via NTP. (Though to ask NIST directly via NTP you need to register with NIST first.) Or maybe you have a common GPS-obtained time source that can serve both the 9037s and STP. Roughly speaking if whatever runs on the z10 could theoretically run at an entirely separate site/company (such as an outsourcer), and the other workloads on the z196s in-house (or vice versa), then this pseudo-sync approach could be viable. You'll then have "second best in class" time stamp accuracy on transaction records across two meta organizations, for example. (Not first best Sysplex, and not third best lesser servers.) "Citibank Eastern Mainframe Time" and "J.P. Morgan Chase Eastern Mainframe Time," metaphorically speaking. -------------------------------------------------------------------------------------------------------- Timothy Sipples GMU VCT Architect Executive (Based in Singapore) E-Mail: sipp...@sg.ibm.com ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
