So why do you need a data center? There needs to be a good business case for having a data center today. If your business is not running data centers, why do you have one - when you can rent computing for cheap.
This is the same thing that a hundred years ago, if you were a manufacturing company you wanted to have your power plant, then came along the utility companies and turned the whole concept on its head. Earlier if you didn't have a power generating plant, you were not competitive. Once the electric power transmission technology matured the companies with captive power plants found themselves with millstones around their necks. Same is happening to the data centers, the data center is akin to the power plants of years bygone.
Cloud is not a marketing buzzword anymore, a mere review of acquisitions and investments being made by IT vendors to know that there is a lot of money riding on this strategy. The IT vendors are positioning for major overhaul in the computing strategy. The switch is on, next couple of years there would be a huge shift.
P B R Trials and Tribulations
In past seven years I have come across (directly and indirectly) five large companies that have embraced Plan Build and Run model. I think much has to be said about that. The attraction of the process is its simplicity, closeness to Demings quality cycle and ITIL framework to some extent. Its graphical elegance has many a companies flocking towards it. In my experience I have seen that execution of this is simple when starting out new. For a company with existing systems and processes this is always a very confusing and trying strategy to implement. Immediately the process, based on my first hand experience, erects glass walls separating infrastructure teams. In absence of good workflow tools, it becomes a major headache to track, who is doing what, and who needs to approve what. The customers initially on the face of it, love the single point of contact with plan team, but soon find out that for delivery other teams are more important than Plan.
Build and Run teams feel marginalized in planning and complain. Us versus them mentality develops almost over night. Planners want to rush away from any kind of delivery support, and "Run" protects its domain by erecting the wall of operational standards. "Plan" plans to meet customer requirements, can't get "Build" and "Run" to participate in design - because typically "it is not their function". As soon as the design is presented a deafening cry can be heard, "Who designed this? This is not standard, and can't be done. Would it not make sense if those who are supposed to support this were included in designing?"
Have seen this happen again and again and again. In one progressive company I saw that the "Run" operation was totally homogenized - analysts were responsible for entire stack. In case of problem raise it to Tier II which happened to be "Build/Plan" team. One can imagine how much sense of insecurity would that have created for those who spent years specializing.
I wonder what is the way to implement PBR? If there is a silver bullet approach? What roles do the leaders play in making this change a success. What kind of change process needs to be embraced to bring the organization onboard this strategy. Thought?
Build and Run teams feel marginalized in planning and complain. Us versus them mentality develops almost over night. Planners want to rush away from any kind of delivery support, and "Run" protects its domain by erecting the wall of operational standards. "Plan" plans to meet customer requirements, can't get "Build" and "Run" to participate in design - because typically "it is not their function". As soon as the design is presented a deafening cry can be heard, "Who designed this? This is not standard, and can't be done. Would it not make sense if those who are supposed to support this
Have seen this happen again and again and again. In one progressive company I saw that the "Run" operation was totally homogenized - analysts were responsible for entire stack. In case of problem raise it to Tier II which happened to be "Build/Plan" team. One can imagine how much sense of insecurity would that have created for those who spent years specializing.
I wonder what is the way to implement PBR? If there is a silver bullet approach? What roles do the leaders play in making this change a success. What kind of change process needs to be embraced to bring the organization onboard this strategy. Thought?
Know-How Dimension
Reading Ram Charan's "Know-How" I started thinking in terms of what is needed to lead an IT organization in twenty first century.
The economy is still stuck in the low gear (Cisco forecast send shudders down the wall street), but innovation in IT seems to be in overdrive. So how do CIOs respond to this new normal? Agility is expected (and is a threshold requirement for IT team), social network integration is imperative to reach the focus market. The vendor partners seem to be finding back doors into IT from the front door. "Cloud computing" has made its way into marketing vocabulary, so would CIO be the catalyst or be the hurdle to it.
The essentials of CIO thought leadership would be - range of altitude, cognitive bandwidth and ability to reposition quickly. The IT leader needs to have the ability to transcend from heights of strategy to depths of tactics. She should be self aware of gaps in her own skills and capabilities to surround herself with the leaders who complement her. With such pace that IT is going through the leader needs to register the clouds of change in the horizon.
What do you think are the essential know-hows for an IT leader? Share your thoughts.
The economy is still stuck in the low gear (Cisco forecast send shudders down the wall street), but innovation in IT seems to be in overdrive. So how do CIOs respond to this new normal? Agility is expected (and is a threshold requirement for IT team), social network integration is imperative to reach the focus market. The vendor partners seem to be finding back doors into IT from the front door. "Cloud computing" has made its way into marketing vocabulary, so would CIO be the catalyst or be the hurdle to it.
The essentials of CIO thought leadership would be - range of altitude, cognitive bandwidth and ability to reposition quickly. The IT leader needs to have the ability to transcend from heights of strategy to depths of tactics. She should be self aware of gaps in her own skills and capabilities to surround herself with the leaders who complement her. With such pace that IT is going through the leader needs to register the clouds of change in the horizon.
What do you think are the essential know-hows for an IT leader? Share your thoughts.
Cloud-Based Infrastructure as a Service Comes to Government
Cloud-Based Infrastructure as a Service Comes to Government
From Apps.gov agencies can access "on demand self service to utilize and discontinue use of products when and as needed. Unlimited storage, monitoring of resource utilization, rapid elasticity, customizable scaling of service, automatic provisioning of virtual machines, storage, chargeback, order management through measured services."
FCC to float cloud computing - Alex Byers - POLITICO.com
FCC to float cloud computing - Alex Byers - POLITICO.com
How big is this news? Per the news article "this will be the first major federal agency to use cloud computing technology to completely power its main web presence".
The website would be hosted on cloud early 2011, to be followed by cloud based storage and email. For security of data the route they are taking is encrypted laptops and VDI.
Enterprises should start developing their cloud strategies, identify low risk data and start testing cloud now.
Fabric based computing infrastructure
Tight integration and sharing of IO resources - data, network, interprocessor, to create virtualized infrastructure in a nutshell is fabric based computing. Cisco UCS is the most publicized example. Obvious benefits "wire once", integrated management, rapid provisioning of compute resources, increased capacity utilization, integration with legacy SAN and LAN infrastructure. High speed network interconnect based on 10gb Ethernet, FCoE provide unified fabric. Migration from legacy compute to UCS would be facilitated by virtualization, vmotion the VM's out of legacy platform to UCS (caveat - source and target CPU's must be compatible). Please share your thoughts on what are the reasons (besides legacy asset life) IT infrastructure managers would not go to fabric based computing.
Storage manager's conundrum
IBM buys XIV, HP snatches 3Par, Dell acquires EqualLogic, IBM buys Netezza, Isilon is part of EMC. All these changes in a short span of 3 - 6 months. If as a storage manager you are scratching your head, you are not alone. This is a hot segment of IT industry. The silver lining in all this is - more options. Technology upheavals from such upstarts have brought breath of fresh air for storage managers. New technologies are fundamentally changing storage paradigms. Thin provisioning, wide striping, virtualization, deduplication, compression, grid processing, ssd, FCoE, pNFS, Hadoop are offering so many choices that didn't exist before
RockMelt - Your Browser. Re-Imagined.
The social network browser. The skin has buttons for facebook, twitter et al. What about mobile devices?
Converged Compute Platforms
Recently I had an opportunity to visit Forsythe Technology Evaluation Center in Skokie, IL. A fabulous facility, a technologist's playground. Equipment from who's who of IT space, which Forsythe maintains and allows its clients to assess before making a decision. I belive the model is neat - companies evaluating technology options can come to the data center, present their test matrix to Forsythe, which configures the test lab, and runs the test. No waiting for the equipment to show up at your own lab, commissioning, configuration, and decommissioning. Think about cost of opportunity savings in reaching a quick decision. Forsythe can help companies with storage, servers, security, network, backup and recovery applications etc.
Strong User Authentication on the Web
A great article addressing authentication challenges on web and especially cloud.
Enterprise Information Architecture
Information is the asset that enterprises harness for creating competitive differentiation.
The development of an EIA should be managed as a formal program by an Enterprise Information Architecture Department with strong participation of business users and clear accountability measures for its success. The EIA must provide the flexibility to accommodate the changes that result from the changing business environment in the organization; consequently, the enterprise information architect plays a key role as an agent of change.
- The Art of Enterprise Information Architecture: A Systems-Based Approach for Unlocking Business Insight
The development of an EIA should be managed as a formal program by an Enterprise Information Architecture Department with strong participation of business users and clear accountability measures for its success. The EIA must provide the flexibility to accommodate the changes that result from the changing business environment in the organization; consequently, the enterprise information architect plays a key role as an agent of change.
- The Art of Enterprise Information Architecture: A Systems-Based Approach for Unlocking Business Insight
Enterprise Architecture Imperatives
Recently I read "The art of enterprise information architecture: A systems based approach for unlocking business insight". There is a trend that most progressive companies are going through for defining their strategies for enterprise architecture. Enterprise architecture is tossed around so much, that it makes sense that before getting into a meaningful conversation to settle on a common meaning of what is this role and what are its competencies. IT focus though correct but as can be imagined is not complete. Earlier focus of architecture was typically make IT efficient, and standardized which are good goals but mean very little to business. IT doesn't define complete enterprise.
Rule of 80 / 20
Pareto's popular theory holds true.
80 percent time spent on 20 percent problems
20 percent time spent on things that would generate 80 percent problems
20 percent employees would produce 80 percent results
80 percent time would be spent with 20 percent of employees who need attention
80 percent time would be spent on things that generate 20 percent results
It is not a bad thing, it is just a statistical reality. What it misses though is the dimension of quality. In organizations that are working on continuous improvement Pareto the "causes" and address them. Once addressed they seek the next top issues.
So don't get disheartened by this number - act on it.
80 percent time spent on 20 percent problems
20 percent time spent on things that would generate 80 percent problems
20 percent employees would produce 80 percent results
80 percent time would be spent with 20 percent of employees who need attention
80 percent time would be spent on things that generate 20 percent results
It is not a bad thing, it is just a statistical reality. What it misses though is the dimension of quality. In organizations that are working on continuous improvement Pareto the "causes" and address them. Once addressed they seek the next top issues.
So don't get disheartened by this number - act on it.
Cloud Computing
Companies can have private clouds if they can provide elasticity, accountability and economies of scale. The challenge is not technical as much as it is financial. Companies can choose to invest in large scale infrastructure deployments to create an infinitely elastic environment - but wait before you do that, what is the cost of doing that? Servers are getting cheapers, hard disks are getting cheaper too, but to deliver a robust compute environment infrastructure one can't ignore the performance and availability attributes. There is where the cost of $ / CPU gets multiplied and $ / GBytes plays a smaller role than $ / IOPS (in a strict infrastructure sense). NetApp, EMC, IBM XIV, 3PAR, HP, provide great solutions for cloud computing, and I would recommending checking their offerings. VMWare has definitely given the organizations the capabilities to think about private clouds. I developed the roadmap for taking a compute infrastructure to private cloud (Infrastructure as a service) in 12 months (post a comment if interested).
Infrastructure is but just one component of the architecture, to implement an application in the cloud, all architecture layers have to be in congruence, starting from enterprise layer all the way to the infrastructure layer. The supporting services need to glue the stack. Risk, compliance, data migration need to be addressed as well. The good (or bad for some) thing is that these issues are addressable at technically, 2010 has seen several technologies that are addressing these quite cheaply also.
The application development groups should adopt practices that would enable applications for the cloud using SOA architecture, office productivity applications like email are easy to move to the cloud (gmail anyone? per seat cost for gmail is way too affordable to pass up). Windows live and google docs, are great examples where companies can move several functions to the cloud, diverting resources to other higher value add tasks.
Infrastructure is but just one component of the architecture, to implement an application in the cloud, all architecture layers have to be in congruence, starting from enterprise layer all the way to the infrastructure layer. The supporting services need to glue the stack. Risk, compliance, data migration need to be addressed as well. The good (or bad for some) thing is that these issues are addressable at technically, 2010 has seen several technologies that are addressing these quite cheaply also.
The application development groups should adopt practices that would enable applications for the cloud using SOA architecture, office productivity applications like email are easy to move to the cloud (gmail anyone? per seat cost for gmail is way too affordable to pass up). Windows live and google docs, are great examples where companies can move several functions to the cloud, diverting resources to other higher value add tasks.
Subscribe to:
Posts (Atom)