Director » Power Systems
IBM announced a new offering on May 26, 2020 called the IBM Power Systems Private Cloud Solution with Dynamic Capacity, which will be generally available on June 30, 2020. While the words “private cloud” are being used a lot today, and while this offering does encompass what the term implies, IBM Power Systems Private Cloud with Dynamic Capacity provides cloud services capabilities beyond just the private cloud connotation.
At a high level, customers place a pool of fully provisioned IBM POWER9 950 or 980 servers, all cores and memory activated, in their datacenter, but with the purchase price based only on a base number of cores and memory activations. This is dictated based on the customer’s usual capacity, not the system’s full capacity. When application use spikes and exceeds the base capacity, variable dynamic use of the system’s fully enabled capacity is automatic, and a utility charge based on per-minute use of the additional cores and memory is incurred.
To minimize costs, the base number of cores and memory activations that you choose and pay for when you purchase or lease the Power Systems 950 or 980 should be close to your average peak usage—perhaps 5% more. You do not have to purchase more cores and memory than you really need just in order to satisfy spikes or temporary growth above the average peak—a number that is usually unpredictable and hence overestimated, increasing system acquisition costs. If your applications use cores and memory close to your base, this will minimize initial costs. IBM Lab Services can help you determine an appropriate base level as a separate consulting engagement.
When you go above your base amount, you only pay for the temporary use of extra cores and memory, which also includes the software charges of AIX, IBM i, Linux, and VIOS. This is similar to public cloud pricing models where there is a base charge and, if you never exceed your base, you incur no additional charge. If you need more compute capacity, it’s there automatically, but you pay for the overage.
Use Case Example
How or why would you use this? Consider an example where you plan to upgrade to three Power Systems 950 servers to support your workloads. These should be distributed in two sites for high availability (HA) and disaster recovery (DR) reasons. The systems use fairly consistent capacity over time, but occasionally businesses experience short-term spikes in system demand, such as during Black Friday sales in the retail industry or when loan rates drop in banking. Let’s assume you are also experiencing 10% year-over-year growth. You determine you need a total of 85 cores based on planned and measured use. With the new IBM Power Systems Private Cloud platform, you might purchase the three 950 servers with 35, 35, and 20 cores as the base on each system, giving you a little more than you need with 90 cores. You place these into an enterprise pool using the new feature code EP2X and IBM activates all the cores and memory so you have 120 cores active in total. That gives you 30 additional cores available for spikes and growth. You could deactivate a dev test partition and keep use under the 90 cores, but with the IBM Power Systems Private Cloud Solution with Dynamic Capacity there is no need. You can just keep everything in place and incur the overage at a per-minute cost, which could be minimal depending on how long the increased demand lasts. This requires no administrative intervention, unlike elastic capacity on demand (eCoD), which accrues a daily on/off charge and requires administrative attention. The IBM Power Systems Private Cloud with Dynamic Capacity offering moves the cost over to OPEX as opposed to CapEx.
The IBM Cloud Management Console, which is included in the offering, gives you visibility into the system and can be used to assess current state, use, and trends. As your business grows and you need more capacity permanently, you can activate more base cores and memory or add a 950 server to the enterprise pool. Plus, you have easy HA/DR enablement.
Does this sound like Power Systems Enterprise Pools (EP) 2.0, which has been around for over a year? What’s different?
IBM Power Systems Private Cloud with Dynamic Capacity is basically EP 2.0 extended with new capabilities, but the concept has not changed. The new offering of extensions to EP 2.0 now supports 950s in addition to 980s but not in the same pool nor does it support Power8 and Power7 servers. The minimum base is one core and minimum memory is 256 GB, which provides a much lower entry point and lower cost. Additionally, you can place the systems in a single pool or distribute them. Support for AIX, Linux, and IBM i (only on the 980) is part of the offering.
A Detailed Progression to Adoption
Here is an example of the steps you would take to adopt the IBM Power Systems Private Cloud Solution with Dynamic Capacity:
- You purchase the system with a base number of memory and cores. There is a premium on activations of these base cores and memory. The minimum base requirement is one core and 256GB RAM.
- On each 950 or 980, you define a base number of cores or memory.
- IBM places a fully configured system (memory/cores) on the floor. If you are unsure about your core and memory use, IBM will install the fully activated system for a month via a Lab Services agreement so you can determine the best base amount for cores/memory.
- You install the systems (minimum requirements of firmware and HMC, 940.1, CMC 1.1, etc.).
- Create the pool on the HMC. The pool can be as low as one Power System and up to 32 systems.
- Register the pool in the cloud management console (CMC). Note that CMC runs only in the IBM Cloud and is required to track the core and memory use of the pool. It needs full-time access to the pool, which some customers may not allow for corporate security reasons, even though it is read only.
- When the CMC initializes the pool, it activates ALL the cores and memory of all the systems in the pool.
- The amount of memory and cores on every system is aggregated across the entire pool as one set of resources.
- You have to set the partitions so they can expand beyond the entitled capacity. If this happens and the aggregate total capacity is exceeded, then the customer is charged a per-minute fee.
- As long as the total pool use of cores and memory does not exceed the total base pool defined, there is no additional charge.
- To pay for the excessive usage, you purchase credits. This is very similar to the pre-pay of eCoD days and requires further discussion but is too detailed to get into here. Mainline can walk you through this.
- If you exceed your pre-purchased credits, your system will still continue to run and you will just accrue charges.
- You can always purchase more base activations as time goes on, as there is a break-even point depending on application use and expansion.
Is IBM Power Systems Private Cloud Solution with Dynamic Capacity for You?
The IBM Power Systems Private Cloud Solution with Dynamic Capacity can be very attractive in many customer environments. Consider this offering when:
- Capacity spikes need to be accommodated
- You encounter sporadic or unforeseen rapid growth
- You want the advantage of using OPEX vs CapEx
- You want to minimize costs of a failover site for HA and DR
- You are hosting multiple users, such as in the managed service provider (MSP) environment
How to Get Started
Contact Mainline to engage IBM Lab Services to help you determine the best base numbers on the system. Power Cloud Rewards can fund this engagement.
The IBM Power Systems Private Cloud Solution with Dynamic Capacity can truly function as a private, on-premise cloud and can utilize Ansible and IBM Cloud Pak for Multicloud Management while running AIX, Linux, and IBM i workloads (980 only).
To learn more about how IBM Power Systems Private Cloud Solution with Dynamic Memory may be beneficial to you, please contact your Mainline Account Executive directly or click here to contact us with any questions..
You may be interested in: