Roy Schestowitz wrote:
> Multi-Core Processors Challenging Traditional Software Pricing Models
>
> ,----[ Quote ]
> | However, Microsoft Corp. hasn't hopped on this train yet; it plans
> | to continue to charge per processor for software
> |
> | [...]
> |
> | Another strategy for CIOs, suggested Giera, would be to consider
> | using open-source software as an alternative to commercial software
> | during the transition period.
> `----
>
> http://biz.yahoo.com/seekingalpha/061009/18110_id.html?.v=1n
This is an interesting article. What is even more interesting is that
IBM has been charging by "capacity units" for about 10 years now. For
example, MQSeries on a single-processor Linux machine, in 1998, might
be one capacity unit. A Windows machine would be 2 capacity units for
the same processor. A PPC processor might be 4 capacity units per
processor.
With LPARs, capacity units are available in 1/16ths of a PPC processor
on PSeries AIX, and 1/4 of a processor on XSeries Linux. Similar
capacity unit pricing is also used for WebSphere, DB2, and Tivoli
products.
The idea is that you can purchase capacity based on the performance
needed, not on how many TCP/IP connections were made, how many "session
beans" are running, or how many transactions are written into the
queue.
On Z-Series machines, Linux VMs can be purchased for as little as
$500/unit. This includes a complement of certain other software.
This is called "On Demand" pricing, and makes it possible for a
customer to set up a system hosted by an IBM Service Delivery Center,
who can charge a lower price for "shared" resources and allocate them
as needed. Customers can by storage for around $10/gigabyte/month, and
Linux capacity as cheap as $500/LPAR per month. Obviously, all of
these prices vary based on quantity discounts and SLA requirements.
IBM also offers "Express" versions, which are frequently used for
development, and are typically limited to a certain number of users,
sessions, or connections. WebSphere Express, for example, can be used
by a department to support up to around 200 users.
|
|
