Udostępnij za pośrednictwem


OpenText: When cash counts, go Microsoft SQL Server

Recently I reviewed Steve Simpson's OPENTEXT document “OpenText eDOCS DM 5.3 - Technical Architecture, Tips and Tricks” and noticed some useful information worth repeating.

Financial Cost difference between Oracle™and SQL Server

I extracted the table below from the above presentation to illustrate key factors emphasized.

Aspect

Oracle Database 11g

SQL Server 2008 R2

Staff DBA

Required

No

Investment Costs

Higher

Lower

Administration Time

High

Low

A dedicated Oracle DBA cost often exceeds $200,000/yr (loaded cost); when this cost is added to a higher cost per transaction for Oracle, we have a money-to-burn solution. SQL Server is documented to have the lowest cost per transaction (as reported by the Transaction Processing Performance Council for the new On-Line Transaction Processing workload). You will see lower costs often cited by users in other posts on this site.

Virtualization of SQL Server

Virtualization is a hot topic and often IT departments see it as the universal solution to budget problems (next year, it will likely be the Cloud). OPENTEXT describes the pro and cons of virtualization with key points, such as:

  • Virtualization may result in higher manpower costs than savings in hardware.
    • Performance tuning becomes obtuse and time consuming
    • Obscurification of hardware reality happens easily. You may discover that part of your database and all of your backups are on the same physical drive (which just failed! Surprised smile)

They sum up the issue of virtualization well:

Virtualize only when it’s going to solve a problem, and you don’t have a better solution for that problem” slide#13

This agrees with my own read on Virtualization for most ISV products. I often use virtualization for code development or tracking down problems by ‘downgrading’ the running environment so blockage and other issues appear. Using virtualization for intensely used enterprise applications is not a path for the meek.