DataCenter Procurement – When to Buy New?

By now, we have seen the building blocks of datacenter like Racks, Cables, Power, Cooling, etc; we have also learnt about its different aspects including how to Managing your Datacenter. It’s time for us to see other components of its life cycle. In this article, we will go through a different view upon planning and procuring datacenter components.

Before we start.. Welcome Back! I have been busy with holiday season, friends and family for a while so couldn’t contribute to my own blog; well its never too late. Okay.. Lets get back to business.

Overview

Most of the organization having in-house datacenter follows a common mission statement: “To Facilitate the Hardware Availability for Data Center Operations & Efficient Utilization, for the Purpose of Business Continuity with less TCO & More ROI.”. I assume most of you know about the DataCenter Life Cycle Management and the activities involved in it. I am not an expert, but I tried to give my view on DataCenter Life Cycle and Management Framework below:

If we focus on the inner most ring of this framework, we can find the core activities in Datacenter Lifecycle i.e.  Analyze, Design, Plan, Procure and Deploy. Analysis happens at very initial stage where we collect information; we hear the Voice of Customer and convert that into requirements. Once you have all the requirements, we need to size the environment accordingly and create a high level design of how things will be deployed or how components will integrate or talk to each other. Later comes the most hectic part of this life cycle. Upper Management, off course, cares about mostly the money. As we have traditionally heard about our AIM for IT Managements:

·         The Reduce the Total Cost of Ownership (TCO)

·         The Increase the Return of Investments (ROI)

Challenge comes, how to achieve it? I have done multiple projects on IT cost optimization in my past experience, I will be sharing one of them in this article.

Applying Lean Six Sigma Methodology

In Lean Six Sigma, I have learnt about problem solving and improvements via DMAIC process i.e. Define Measure Analysis Improve and Control. Now going back to the problem that we all are facing with High Cost, I tried to use Fish Bone technique to give an overview of how we are getting High Cost in IT Operations in DataCenter. Below is my very basic analysis of high cost, off course, there are multiple branches, pros and cons for every consideration, but like I said, this is just an overview to help you understand where I am coming from.

If you notice, under Money; Annual Maintenance Cost (support) and Procurement are the two major contributors to Operational Expenses and Money expenditure for every environment. Now, I am sure you might have head How Virtualization makes your life easy with features to bring in like Centralize Management, Less support cost etc, so I am not touching on that. However, Lets talks about how we can reduce cost during procurement itself considering virtualization and other techniques.

Legendary Procurement Process “As Is”

Below is the “As is” procurement workflow followed in usually in every organization.

 Now, as you can see there are multiple flaws with this process:

•  There is No check on optimized usage of existing hardware before new procurement

•    There is No check on timelines of usage of New hardware to be procured.

•    There are No Guidelines available to choose the Right Hardware for project.

•    There is No check if Virtualization can avoid procurement

•    Methodology to Track the current role, usage & future plan of Data center hardware is Missing

•    Unplanned procurement leads to wastage of CAPEX & OPEX Budget.

Now to make it better, we need first put in our thoughts to Identify the need – what you really want?

• Do I need More Hardware in my Environment?
• Does present solution meet my needs?
• Do I have Areas  of Improvement?
• Do I have 100% Utilization of Existing Resources?
• Am I buying Hardware as per the Roadmap?
• Do I need to Invest to Manage existing Resources?
• Do I need Business Continuity or Data Availability?
• Do I have enough Hardware to Meet the Dynamic Needs?
• Do I Track the Usage of Existing Hardware?

New Approach

Definitely, before we proceed with procurement, we need to identify what we have and analyze what we need. Now, let me introduce you a new way of procurement:

You can call it a process or decision tree for “When to Buy”, whichever way its easy for you to integrate with your own environment. I understand its not easy to Bring Change in an organization but when people have common motive, its easy to drive atleast. As you can notice, it highlights the checkpoints where you can think – Can I Avoid Procurement?

• Check if free hardware is available
• Check if hardware require immediate deployment
• Check if free hardware is available during required timeline
• Check if free hardware is available in other section, Biz-IT or Partners
• Check if server required can be virtual server, if No, Why? verify from TC
• Check if storage connectivity would be required.
• Check if dedicated path is required for Server
• If No, Raw Device Mapping with Virtual Machine can be used
• If Yes, VMDirectPath can be used (VMware Technology, similar is available for Microsoft Hyper-v as well)
• Check if Simulators can be used

To understand it better lets highlight the changes into sections:

1.   The Asset Management - Inventory and Audits: collect your existing hardware inventory and perform audits for its usage and health. This will help in doing projections and planning well in advance.

2.   Usage Tracking: With available reports, find out if there is any free hardware or capacity available for deployment. Usually DataCenter Life Cycle Managers have this information.

3.   Shared Resources: At times the resources can be shared or can be released in case of Test & Dev environment. Another example is of Public Cloud Hosting.

4.  Virtualize: Check the possibility if it can be virtualized. I consider most of the environment already have virtual infrastructure which may or may not have capacity for hosting new VMs. In case you don’t have capacity, rather investing in buy new physical server, invest in increasing the capacity for existing virtual infra and share its resources. You can also go Hosting services these days offered by many vendors.

Well, that was just an overview, but there are many aspects of saving cost. However, to start with, this approach might save some initial investment itself. Why to optimize later when you can cut cost in initial stage, right?

Any more questions? please write back or comment here. There are more things to share.. 

Request you to join my group on Facebook & LinkedIN with name "DataCenterPro" to get regular updates. I am also available on Tweeter as @_anubhavjain. Shortly I am going to launch my own YouTube channel for free training videos on different technologies as well. 

Happy Learning!!

DataCenter 101 - Documenting DataCenter - Make it Visible

By now we had an Overview of DataCenter; we learnt about Racks, Cables and its labeling too. We will go in-depth about few concepts for Power, Cooling, and HVAC in future along with learning technologies like Blade Enclosures, Virtualization and SAN Storage. But first, since now you might notices what you have in your datacenter, lets capture it first. Documenting your datacenter is the most important phase in your datacenter deployment. It’s important when it comes to Managing your Datacenter. 

I have seen people capturing Server Implementation Plans (SIP), Network Implementation Plans (NIP) and Standard Operation Procedures (SOP) for New Deployments, but where is the mapping of this information to actual physical deployments. You may an inventory available with you where you have your rack location against your server/storage/switch equipment, but can visualize your datacenter with that? In a way that I give you the information in excel sheet, and you are new to your company and something goes wrong; now I need your assistance to walk-in to the datacenter and check if the cables connections are fine for a particular setup; so you think you can do that with your current available information? Imagine the time you would lose in tracking it even though you have labels in your datacenter. You know time=$$$ right!! So I suggest, invest sometime in making it how you visual it so that you don’t have single point of failure in absence of people who deployed it.

I believe you are convinced by now that why you need to create datacenter documentation. So let’s talk about what need to be captured (suggested) in this document and what you need for that.

To Be Captured:

1.       Overview: It’s like a summary of your datacenter, how it looks. I prefer to mention who designed the document and who owns the document along with who reviewed and approved the document as important information to be mentioned in the beginning itself. It’s give a level of confidence on the document. Like I mentioned, this shows how it looks don’t jumble the position of equipment, racks and their location within. It should be like a photo as it looks.

2.       Power: I believe it’s a good practice to mention the current operating power and max requirement specification based on engineering design. You would need assistance of your local maintenance guyz to measure operating power in peak hours. Suggest you to take max noted in one week or you can take average as well. As per my experience, usually 3 phased power is used in a datacenter. In this example, R-Red, B-Blue and Y-Yellow is named as three phases and its consumption across a rack. You need to notice on the total power available in a circuit and its consumption.

3.       Network LAN: Next comes the Network connections, ISL and bandwidth usage presentation. I would suggest to make one more sheet to represent the IP subnet used in individual racks along with number of ports. This comes handy when doing a root cause analysis in case of network break-down.

4.       Storage SAN: Storage connections are usually tricky and gets messy if your datacenter goes scalable on-demand without proper planning. It is suggested to give a visualization of how storage connections are made across racks and then a logical presentation of ISL connections. Its comes handy when troubleshooting performance issues.

5.       Server Connections: I always to show case a cable connection within a rack and it should be pasted or available as reference material with critical deployments. Imagine you had to remove cable connection for maintenance or replacement of any components and cables get swapped? This is suggested for deployments in a rack.

6.       Setup Configuration: This is a brick level logical presentation of your deployments; where you map how your physical infra (Server, Storage, Network and most important Applications) exist in different layers. I have seen this in Server Implementation Plans (SIP) but not sure if everyone provide the detailed information about the deployments every-time like physical rack information, Management IP etc. Sharing some examples below:

7.       Contact Information (Disaster Run-Book): This section should always come either first or last, preferred last though. This should mention the on-call IT team members name and contact information along with respective areas of expertise like Windows, VMware, Server, Storage, Network or domain of ownership like Power, Cooling, Space etc. This comes handy and becomes a reference in case of Disaster. You can also mention Vendor support information like Contract IDs, Support Numbers, Account Managers Info etc so that you don’t have to search it everytime.

What You Need:

8.  Visio: When it comes to tools, Microsoft Visio is the best one I have seen. You would need stencils to give a real picture of your datacenter. Generic ones should be avoided or should be used only for logical explanations like routing/firewall setups, but what if you need to show actual physical connections? I have seen many usual stencils available on Visio-cafe (ask Google); rest you can ask the vendor for it, they usually provide it as free of cost. Note that there are certain open-source tools available, but they may or may not give the precision to show-case your datacenter. I may be biased, but I am just a fan of Visio.

9.   Information: Apart from tools information is most important. It may be available like a document for setups or like an excel sheet for inventory. Make sure you validate the inventory before you actually deploy it. Most important, you need to make sure this information is kept up to date.

10. Owner, Reviewers & Approvers: To make sure you have the proper information and its mapped as is; Get it reviewed from your peers (Important). See if they have any comment or suggestion or doubts before you actually publish it. What’s the use of it, if they don’t understand it. Like I mentioned before, its less technical document but more logical document. Owners are required to keep information up to date and approvers are required so that this can be use a standard reference documents and can be published.

Special Notes:

There are couple of tools available which easy your work for making automatic diagram via environment discovery like HP SAN Visibility (free), VMware Maps (free) and some other third party tools. However, you would need to consider the pros and cons that comes with these tools like support limitation, cost, credentials or access requirements etc. Adding to it no tool shows you rack placement.

Any more questions? please write back or comment here. There are more things to share.. 

Request you to join my group on Facebook & LinkedIN with name "DataCenterPro" to get regular updates. I am also available on Tweeter as @_anubhavjain. Shortly I am going to launch my own YouTube channel for free training videos on different technologies as well. 

Happy Learning!!

P.S.: I understand the example was given using small deployments, but it may look tedious when it comes to bigger deployments. Yes! It will be, if done by one man army J suggested to divide the work with respective areas of expertise or domain to make it easy, which can be later complied together.