No Room For Cloudy Thinking When Moving To The Cloud
October 15, 2018 Doug Mewmaw
With new IT innovation, comes new IT thinking. That’s the way it’s always been.
Right now, that new shiny IT toy that everyone must have is the cloud. In fact, the global market for managed services will grow from $107 billion in 2014 to $193 billion by 2019. As a result, Managed Services Providers (MSPs) are trying to convince CIOs to not only outsource their data center administration, but to trust them that they can manage their entire IT enterprise cheaper and more efficiently.
Unfortunately, they both may just have their head in the clouds. You see, I’ve been here a few times in my IT career. Case in point, I remember in the mid-1990s when all the CIOs decided that we had to put all the processing power “out with the people.” In fact, my old company began building regional call centers everywhere. These were gorgeous state of the art buildings, and I remember how happy everyone was when the new AS/400s went live. I’m not here to debate the pros and cons of distributed processing. I do remember the shock that occurred when major issues arose on these remote systems and the realization all the people that had the expertise to fix these systems were thousands of miles away. In other words, the call center director was a great guy, but he didn’t have a clue on how to manage the IBM i infrastructure. It was a teaching moment for everyone.
Back to the cloud discussion and my déjà vu that is occurring every day in regard to what is happening in our industry currently. I have visited 200-plus customer locations in the past 14 years, so I feel I understand the pulse of what is occurring within the sys admin world. In the past year alone, I have visited three customer locations that have made the decision to outsource the hardware administration processes to an MSP. I have also visited two MSPs. We have had some interesting discussions. As an ex system administrator, I want to be perfectly clear I have no issues with companies making the decision to have someone else manage the complexities of the hardware & software within their enterprise. However, I would be remiss if I didn’t stress how important it is to learn from our past.
Let’s look at a simple scenario:
- A Power8 customer has decided to move to a new Power9 system.
- The CIO has decided to have a MSP build the new hardware configuration as well as manage all related hardware processes (backups, PTFs, OS upgrades. . . .).
- The company is also upgrading their warehouse automation application which will run on the new Power9 hardware.
The system admins that have managed the hardware for years, will provide the business partner the current hardware specifications to help with the new capacity plan. By providing an enterprise hardware summary of resources, like this:
During the planning phase, the CIO and all system administrators must ask the following questions:
- What is the impact of moving the enterprise to a managed service provider?
- Am I on my own LPAR or am I sharing resources with others?
- Which brings up the most important question: What are my service level agreements? (Response time, guaranteed CPW, disk response time, and so forth)
- What is the actual hardware that we are moving to? (number of cores, clock speed, disk configuration, and so on)
MSPs must remember the golden rule for customers: It is not just about capacity planning and providing hardware.
If you think about it, just because the hardware is no longer in the customer’s building, doesn’t mean performance issues are going to stop occurring. In fact, the MSP/customer relationship becomes even more complex.
Which brings up even more questions to ask the MSP:
- What are the processes when there is a performance issue?
- Who is my MSP contact?
- Can I call 24×7, every day of the week, every week of the year?
Here is the scenario that MSPs are forgetting (in fairness, they are hardware providers first). Customers have performance management software running on their LPARs (for example, HelpSystem’s Performance Navigator, IBM’s iDoctor, and others) to do mission critical performance analysis and reporting.
The questions to the MSP need to be: Will you be installing the software too? If not? Why not? The CIO is not going to stop asking a slew of questions, such as:
- What’s wrong with the system?
- What happened yesterday at 4 PM?
- Why is disk space growing at an alarming rate?
- Why did night batch processing run long last night?
- Are we going have enough processing power for our peak period?
And so on.
Here is the bottom line. Because customers can have so many issues, MSPs must create an IT culture where customers can access their performance management data easily. It allows customers to do their structured performance reporting, to do problem determination, and to do capacity planning. The only other alternative is that MSP’s must train their associates on all the mission critical performance management software. The MSP would have to create a customer liaison process to communicate back and forth when performance issues occur.
Here is the bad news for the MSP. Both scenarios would require the hardware vendor to purchase performance management software licenses to not only service the customer effectively, but to be compliant with the applicable software vendors.
Moving IT hardware and application software to the cloud is really a cool concept. With that said, I hope you can see this project is a lot more complex than calling up a vendor who has a bunch of LPAR space and CPW capacity available. If you’ve been assigned this project, take your time and really understand the impact of moving the hardware out of your building. My best practice suggestions would be as follows:
- Ask a lot of questions.
- Don’t assume your hardware vendor is providing you with the exact configuration you had initially.
- Have clearly defined SLAs that work for your company’s environment.
- Measure your core metrics on your original system before the cut over.
- Measure your core metrics after you go live on the MSPs system.
- Compare the performance to make sure you are getting bang for your buck.
Like I said, let’s simply learn the lessons from our past.
RELATED STORIES
The Cloud Changes Things, But Not Everything
A Better Way To Skin The IBM i Cloud Cat
Google Close to Launching IBM i Cloud Service, Will Says
Good article, but you forgot one important thing – Audits. Even though you have moved to the cloud, you, not the MSP, are still ultimately responsible for audit compliance (ILMT for IBM PVU based software, SAP, Oracle, etc).