This is part 3 of the series “10 Things Vmware Server Admins Should Know About Self-Service Catalogs and Lifecycle Management” that I’ll be publishing over the next couple of weeks.
3. The catalog system is more than a document, it’s also used to manage the life-cycle of the resource
What’s great about VM’s is how fast and easy they are to provision, but sometimes they are hard to kill.
I see the emails going around that say: “no one is touched that instance, who owns it?”
Back when resources were scarce, our hunter gatherer customers in Application Development and QA learned to never let go of a server. Like woolly mammoth’s they were hard to catch and came only sporadically; in the summer of ROI funding, or when great migrations came. Most of the time, QA was starved for resources. So they hoarded.
And while executing the initial request for a server environment through the service catalog gives you a nicely documentation and speed, over time changes happen and configurations drift.
This process of managing a server or environment from “as offered,” to “as agreed,” to “as built,” and then managing the change requests against it, is what I mean by lifecycle management.
The service catalog, being the source of “as offered,” “as requested” and “as built,” contains the whole lifecycle for your VM, plus information on who owns it, for how long they need it, and any other relevant data that went into the build sheet.
Unlike a static spreadsheet, when looking at a server, you can see what the maintenance hours, SLA’s and OLA’s are. The lifecycle system an tell you what types of requests can be made against that VM (like add memory, for example).That server can be started, stopped, snapshotted, upgraded. Notice they are all verbs against a thing, the VM instance.
The result is we have complete business context information about the server, the history of requests about it, subscription information and of course the proper technical build sheet, including workload requirements. As one VMware admin recently said, “I wish I’d known that you can only work on that server on Saturdays after 5pm.”
If you’re like me, you’ve ignored this sage advice a time or so. Thankfully my most recent rush to a solution was remedied by a trip back to the home improvement store and $100 or so. Most IT “goofs” extract a much dearer cost. In this third installment of Cisco Insights – Cloud, Bob Dimicco profiles a non-profit company who successfully resisted the pressure to jump straight on the Cloud project bandwagon. Instead they opted for a thoughtful, measured approach. Partnering with Cisco, they first conduced a thorough strategy and business justification assessment. By focusing on their key business drivers and desired outcomes, they were able to get a complete picture of benefits, costs, and a deep understanding of where the true ROI would be. They allowed the facts, not the hype or pressure to guide their direction. Watch the video find out where they went next on their journey to cloud.
Read about two other companies’ cloud stories in my previous two blogs:
Leading IT shops like to have a single pane of glass that is the IT storefront to all employees. This is a very noble goal. Having worked at a few large companies this is indeed a moving target as supporting the end user employee can mean a lot of different entry points, contexts and presentation technologies. When it comes to have a central location for ordering services it is very important to on board all of the employee based and data center services in a consistent fashion. Some of the key use cases include employee on boarding (and off boarding), virtual desktops, virtual machines and physical servers in the datacenter and access to applications. Typical IT departments may have several hundred orderable services, many of which are bundled (think of employee on boarding).
Interestingly some organizations first drive towards a common catalog and then automate what they can afterwards. At first you can take orders through the service catalog and then work the tasks to fulfill the request through manual process tracking. Alternatively I have seen some shops say that they will only put services in the catalog that can be automated. Then there are all the intermediate cases. Organizations deploying automated request management have many issues to consider and standards to be set.
Can we declare victory when a process is mostly manual but yet orderable from a catalog in four clicks? Perhaps…
Your end users are happy. They can see where their request is in the process flow. Kind of like going to fedex.com and seeing where that DVD is on its journey to your house. But that package took 3 days to traverse its journey.
Considered an automated fulfillment or provisioning process. In my above analogy, you are no longer dealing with DVDs shipping to your house but on demand video streaming. A simple click sets into motion many automated processes that deliver the movie to your device. For end user services this means your remote access is provisioned with a simple click, your Linux server and application stack is delivered in less than 15 minutes for use. Key to making that happen is a full automated process. Is that achievable in all cases? Perhaps….
In most cases what we are provisioning requires a northbound API (an programming interface above the fulfillment system) to accomplish the instantiation of the service. Oftentimes, in legacy environments the target system is so dated or under invested-in that an API does not exist. It is pretty hard to automate a process that can only occur through a human interfacing with the system.
People ask me the question: So What? We have found that by automating processes we can save on average 30% of the process cost. Multiply that by tens of thousands of requests and it will really add up.
Investing in Self Service requires investing in automation and in some cases, wrapping an API around a legacy environment in order to get the desire result: IT as a Service, delivered at the speeds needed by our end users.
Its true, there is no rest for the weary. While we are putting the finishing touches on CiscoLive!, we have already started work on VMworld 2012, which remains one of my favorite shows. As part of that, we have submitted the following session topics for consideration.
1988 - From Here to There: VMotion Within and Beyond the Data (by yours truly): One of the coolest aspects of vSphere is VMotion. There are a number of innovative technologies available to help you make the most of this feature. This session will help you understand the use of various technology options such as flat architectures, VXLAN, OTV and LISP to support VMotion within the data center and between data centers. As with many aspects of IT, there is no one right answer. The session will discuss the pros and cons of various technologies to allow you to decide what best meets your needs. And, since no VM is an island, the session will also look at how L4-7 and storage figure into things.
2680 -- Secure multi-tenant data center with Cisco ASA1000V, Virtual Security Gateway and Nexus 1000V: Cisco ASA1000V cloud firewall and VSG together provide a comprehensive cloud security solution. The Cisco ASA 1000V Cloud Firewall employs mainstream, proven Adaptive Security Appliance (ASA) technology, optimized for highly secure multi-tenant virtual and cloud infrastructure at the edge. Implementing Cisco Virtual Security Gateway (VSG) with the Cisco ASA 1000V Cloud Firewall in a virtual multi-tenant data center solution provides tenant edge, intra-tenant, and inter-tenant virtual and cloud security.
2373 -- Best Practice for Deploying VXLAN with Cisco Nexus 1000V and VMware vCloud Director: Cisco Nexus 1000V is the first virtual switch to provide Virtual eXtensible Local Area Network (VXLAN) providing cloud network isolation and is fully integrated with VMware vCloud Director. Come to this session to find out how to deploy VXLAN with VMware vCloud Director. This session will also provide a deep dive into VXLAN deployment best practice.
2227 -- Go Big! 10G and Multi Adapter vMotion for Large Workloads: The addition of Multi-Adapter vMotion and improved overall vMotion performance in ESX 5 allowed Medtronic to scale up to a high density virtualization and large workload environment with 1TB of RAM per host and guests with up to 256GB of ram. This session will explore the networking challenges and solutions of “scaled up” virtualization environments including the configuration of multi-adapter vMotion, NIOC, class based WFQ QOS in the Cisco Nexus 1000v, and QOS in the Cisco UCS fabric.
2352 -- CSC Case Study: 10 Weeks to a Private Cloud: In this session, learn how leading global business solutions provider, CSC, is providing innovative private cloud services to its clients through its new CSC BizCloud – the industry’s only opex private cloud that is billed from a standard rate card and ready for workload in just 10 weeks. Using the Cisco Intelligent Automation for Cloud software solution running on Vblock infrastructure – along with VMware solutions – CSC BizCloud provides users with a self-service portal where they can order services, but also manage access and track usage to manage costs. This solution orchestrates workflows across technologies, so CSC can have the flexibility to provide new self-service options to clients over time.
2541 -- Case Study: Bank Increases Agility with Private Cloud: Intesa Sanpaolo turned to Cisco and VMware to help simplify its IT operations with its Next-Generation Data Center project – which delivered IT services in a private cloud model. Using Cisco Unified Computing System and Cisco Intelligent Automation for Cloud software – along with vCloud Director – the bank was able to integrate with existing infrastructure while adding new systems into its data center. These products provided easy self-service provisioning and access to resources, plus automated orchestration of infrastructure. The benefits were simplicity, configuration flexibility, easier enforcement of governance policies and technical standards. As a result, provisioning times have been shortened from weeks to minutes.
This part 1 of the series “10 Things Vmware Server Admins Should Know About Self-Service Catalogs and Lifecycle Management” that I’ll be publishing over the next few weeks--I hope! (The boy is nothing if not ambitious).
1. The service catalog is a tool for driving users to standard configurations.
To get the operational efficiencies we hope to achieve from virtualization and / or cloud computing, we need to establish standard configurations. This is tough, for a couple of reasons.
First, the challenge is the gap between the language of the customer, and the detail needed by the operations group typically generates a lot of back and forth during the “server engineering” process. Instead of having “pre-packaged” configurations, every thing is bespoke.
Instead of having useful abstraction layers and levels, the customer has to invent their own little bit of the data center. This made sense when the new app meant a whole new hardware stack to which the app would be fused to and the concrete poured on it. It doesn’t make sense now.
Second, there’s resistance from customers to adopt standard VM builds. Sometimes the reasons are valid, other times less so. The issue arises because the technical configurations have not been abstracted to a level the user can understand what they get and what’s available for configuration. Nor can they compare one template to another in ways that are meaningful to them.
The service catalog is the tool to help deal with these two obstacles. The service catalog is a useful tool to communicate, in the language of the customer, the different options available from IT for hosting environments.
A service catalog will support multiple views (customer, technical, financial, etc) so that when the customer selects “small Linux” for testing, this generates both a bill of materials and standard configuration options. Once that base is selected, self-service configuration wizards provide both guidance and gutter-rails so the customer is both helped to the right thing and prevented from making errors.
From this customer configuration, the environment build sheet is generated which will drive provisioning and configuration activities or to execute any policy automation in place.
And the catalog allows the VM admins to figure out what their “market” is buying; which is very useful for capacity planning.