Re: Version blues
Date: Wed, 10 Jun 2009 06:36:50 -0400
Bob Jones wrote:
>> Where did I mention shared nothing?
> GridScale IS a "share nothing" approach. Each instance in the cluster has
> it's own database.
Typically people refer to DB2's DPF as shared nothing, but OK. Technically xkoto is as "shared nothing" as Dataguard or replication. How is that a bad thing?
>> The relationship between IBM and xkoto is pretty tight. They are literally >> only a subway token away
> Regardless of how tight the relationship is, GridScale runs outside of DB2
> and is licensed separately. They are likely deployed in separate tiers as
How is that a bad thing?
>> xKoto basically drives two (or more) replicas of the database. >> (similar to a RAC or Dataguard assuming you have true HA on your disk).
> A single update will need to be applied on each cluster node to keep data in
> sync. That's more similar to replication not RAC.
It solves the same business case. Isn't THAT what you woudl define as "similar"? If you want DR in RAC you need to 'double up" with Dataguard which gets really pricey....
>> HADR (somewhat similar to Dataguard) gives you HA with few seconds >> failover time and near zero extra admin skills needed. That's used for >> most SAP systems on DB2.
> Again, that is not RAC.
Again why is that bad? It does the job, it's cheap what more can you ask for?
What would be a "similar" solution in your world? Does a DB2 Sysplex count as similar to RAC?
PS: On July 16 I give talk on
"Enabling applications from Oracle to DB2 the easy way" at ibm.com/db2/labchats
11:30 AM Eastern / 10:30 AM Central / 8:30 AM Pacific / 4:30 PM London / 5:30 PM Munich
Would love to hear your thoughts in the Q&A afterwards (or before)
-- Serge Rielau SQL Architect DB2 for LUW IBM Toronto LabReceived on Wed Jun 10 2009 - 05:36:50 CDT