OPS scalability in 8i? thoughts?

From: Sean Emery <emerys_at_remove_me_jagunet.com>
Date: Fri, 15 Dec 2000 00:03:48 -0500
Message-ID: <5C66450B4EF2B7E0.6B6C5206B44DA827.2EAB307E38880C2E@lp.airnews.net>

Hello,

I've been wondering lately how well the newer versions of OPS - specifically 8i and later - scale w.r.t. OLTP applications that may not happen to be specifically tuned for OPS. Does cache fusion help here and, if so, how much? Is this quantifiable?

I sort of expect the answer to the above to be "don't use for generic apps" on concurrent access to multiple instances, but how about for DSS apps? 8i cache fusion should help here, no? Anyone have any in-depth experience with either of these apps and care to share impressions? Hardware recommendations? How exotic have you gotten with node interconnects (please include max throughput)? Is it important to invest in better than ethernet
(100base-T)?

In addition to scalability, I think many shops are considering OPS for HA.
(Oracle is certainly touting it for such.) Can anyone give some impression
of the amount of "extra" work OPS might be in this configuration if you only use the "secondary" instance for failover? Should be minimal extra tuning involved in this case, any gotchas?

I've been working with Oracle for some time now (4 years), but have no hands-on with OPS yet. Thanks in advance for any insight!

Sean Emery Received on Thu Dec 14 2000 - 23:03:48 CST