Flirting with Cisco UCS (Unified Computing System)

i recently got an opportunity to work for a customer’s DC consolidation requirement where we Cisco’s UCS was explored as an option.

this is a tagholder for the post 🙂 – i will post my analysis along with some numbers to present my experience with Cisco UCS soon.

Advertisements

Clouds & Infrastructure Management

recently, someone asked me a question around infrastructure management – One of our business units are going in for cloud computing. Do they still have to look at infrastructure management?

 

here are my thoughts on the subject –

 

Typically infrastructure management involves design & planning of infrastructure components, procurement, deployment, operations and disposal. However, cloud computing introduces a different aspect – cloud management activities.

Cloud computing has various flavours like SaaS, PaaS, IaaS from services point of view. Also from ownership model, there are public clouds which are owned by service providers and then there are private clouds which offer cloud computing services but are owned by the customer themselves and reside in their data centers.

Depending on which service viewpoint and what ownership model is the customer considering cloud computing, there will be different levels of involvement in infrastructure management. Public cloud providers have solutions that involve a lot of automation thus making the infrastructure management activities simplified.

for public clouds

SaaS service does not require infrastructure management PaaS service does not require hardware level infrastructure management but will involve activities like provisioning and de-provisioning (cloud management activities), patch management etc.

IaaS service will require activities like provisioning, de-provisioning of compute resources, OS & application patch management etc.

for private clouds –

PaaS & IaaS will require infrastructure management like provisioning of hardware for setting up of private cloud and then cloud management activities like provisioning and de-provisioning of compute resources (cloud management activities), OS and platform patch management, incident, change, problem management etc.

 

from email to collaboration

initially i had titled this post as “email and its imminent demise” but then maybe it is incorrect to say “imminent demise”, rather i think it will slowly move into the background and slowly treated as legacy application if it fails to evolve and incorporates the new web 2.0 technologies that ppl are toying for enterprise use..

email has come a long way since it was invented in early 1970’s. people have used emails to  send simple messages to each other/group, use it as means of sharing documents (much to the plight of the email administrators), use calendaring features for appointments and scheduling meetings. on looking closely at how an email is typically used in an organization, i have noticed that along with the above mentioned functions,  emails are also used as means to store files (hey dude, can u email me that presentation that you took for a xxx client the other day, i may use it in future),  approvals for certain business transactions (e.g approval to buy a new server sent by CIO to IT Manager etc), keep a record of certain communication (he said, i said etc) to basically CYA (covery your a**) in case things go bad..however, along with email came its nuances, maintaining uptodate address book, spam, compliance issues etc. but we are all dealing/living with it.

as time goes by, in the world outside the boundaries of an enterprise, people are adopting techniques that harness the power of technology to reach out to each other, either as part of social networking phenomenon or to work together/collaborate. with the power of the internet coming to the mobile device, the speed at which this adoption is taking place is awesome. now, people dont need to access their emails to know what their friends/peers are doing or what is the latest buzz etc. applications like facebook and twitter have already proven the usefulness of technology outside the enterprise walls..it is only a matter of time when they are adopted within the enterprise also. already there is a lot of talk about web 2.0 adoption within the enterprise by many analysts. email too needs to evolve if it has to stay alive in this fast changing scenario. already there are surveys which shows that more and more people are using facebook/myspace/twitter etc over email to reach out/keep in touch.

in my opinion, organizations will start looking for a better collaboration platform which can increase the effectiveness, efficiency & productivity by harnessing the same technologies that are used by millions outside the enterprise boundaries. realizing this need, some email services have now integrated instant messaging/chat services with archiving features on the messaging platform. . how many times have you noticed that when ur having an email exchange with someone, u end up taking the remaining conversation to the chat as it is faster and more effective. i think the collaboration tools like instant messaging will  evolve and bring convergence of channels of communications viz chat, voice, video along with email and other forms of collaboration technologies like document management systems etc. the other day i wanted to reach out to a group within my company to work on platform migration. instead of sending emails to numerous people and then getting redirected from one group to another, i just sent an update on yammer and got a reply within 2 hours from someone located on the other side of the world.. it was awesome and saved me numerous emails and days in waiting for a favorable response.

there will be initial resistance from many to move from legacy to new forms of collaboration. there will be compliance & security concerns. but like the case with any new technology, it will find its own steady state adoption rate. only that, given the way web 2.0 is evolving,  this rate might be accelerated by few notches.

already there is a lot of noise that google wave has generated on the internet. maybe microsoft solution around corporate instant messaging – OCS will also evolve to bring in convergence of existing communication & collaboration tools with the new web 2.0 toys.

my take is that few years from now, email might just end up as a legacy platform required to retrieve old email data and corporates will use a newer and more efficient form of collaboration technologies.

Cloud for IT Continuity

typically a DR site goes live when the main DC goes offline of fails. quite often, the IT infrastructure at the DR site sits idle waiting for an untoward incident to be kicked back into life. in some cases, the infrastructure at DR site is used to host dev & QA environments also. the DR sites are typically activated for a short period of time and when the main site/DC is restored, the DR goes back to idle state. is there an alternative to blocking investments in a DR site using the evolution in the technologies used in DC and still ensure continuity of operations?

can cloud & cloud based services provide enterprise with the desired level of continuity along with financial flexibility? in my opinion, this is a subject worth further exploration.

during a disaster, you either operate at same or reduced business service SLA’s around performance & availability as from the main site.  the requirements from the DR site are “elastic” in nature,  most of the times, the compute requirement around CPU, memory are pretty low except when activated and operations are run from the DR site.  usually it is the storage that has a consistent use. now, one of the major advantages of cloud computing is to meet elastic demands. put two and two together..i feel there has to be a case to use cloud for IT continuity!

one of the possible challenges is the consistency of the virtualization technology within the enterprise with that of the cloud computing provider. i do not think the cloud computing providers fraternity has something of an intera-operable virtualized images across different cloud providers and private cloud platforms..(or maybe they have. this is something i have not tracked in the google-sphere yet!). so basically what that means is you are stuck with those set of cloud computing providers who use the same virtualization technology as you use in-house in your DC’s for the time being. but compared to having idle investment in your dedicated DR sites, this may be a small trade-off.

some points that i can think of while evaluating the cloud platforms for DR & IT service continuity is – licensing of your existing apps..does the licensing allow you to run the apps from a cloud computing setup, connectivity options to allow migration of large amount of data/images to the cloud computing provider’s setup, how are you going to keep the images of your apps etc in the cloud environment up-to-date with necessary patches, security policies of the providers and client access mechanism.

will update as and when i have discussions with more customers on this topic!

Random Notes on Cloud Computing!

this post captures some random notes i have come across & my thoughts on technical aspects that can facilitate the cloud computing environment. these are not in a structured order, so bear with me!

1. Cloud computing is a way to maximize capacity and utilization and to minimize space, maintenance and to simplify governance.

my thoughts – Does it actually simplify governance is something that is yet to be seen as governance also encompasses security, risk & compliance along with service orchestration.

2. Virtualization is not a cloud solution, but a cloud solution will require virtualization in some form, whether it be cloning or full virtual images.

3. Parallel processing on pooled resources is not a cloud but the principles of that are important to the conception of an effective cloud.

my thoughts – absolutely in agreement with point 2 & 3

4. A cloud also requires understanding of the enterprise, a clear picture of patterns and topologies and an efficient process for managing images as distinct entities.

my thoughts – Cloud computing will have an impact on the Enterprise Architecture of an organization to address the new patterns and topologies.

5. Cloud bursting –  The scale out should not require tremendous effort & specialized skills otherwise the benefit of cloud computing may be lost or reduced.

my thoughts – cloud bursting requires a thorough understanding not only to move from private to public cloud but also enable the reverse. i agree with IBM on the point that applications hosted in the cloud need to run on same platforms as enterprise applications to facilitate movement between the enterprise and public clouds. not everthing can be free in life 😉

6. Scaling out for Scalability – Running another instance of the application on another server(s)

my thoughts – Typical scalability in the cloud is provided by scaling out and not scaling up in cloud computing frameworks. this will also depend on the way the application logic has been written to benefit from multithreading, multicore and multiprocessing technologies that are/will be available in the cloud. the way an application logic is written will eventually determine the ability of the application to seamlessly scale across multi cores, across physical servers and be able to withstand & survive any infrastructure failure.

7. Load balancing – balancing the work across multiple systems in the cloud

my thoughts – usually most of the cloud players will allow you to create exact replica of your systems thus balancing the transactions across these set of “clone” systems. if using atomic code, one can also allocate specific systems for specific tasks instead of creating the clone of the entire application system.

 

8. manageability – ability to manage the cloud systems seamless with lower management overheads

my thoughts – management of the cloud based systems will be a become a big ticket item in times to come for enterprise and cloud providers, both. this can be achieved by using virtualized systems and layer of automation to ease the provisioning and de-provisioning of resources on demand. enterprise will look at how cloud will deal with the applications to be deployed. using the process of cloning of systems, enterprise will prefer multiple instances of applications can be implemented with few clicks of the mouse instead of deploying the application on each virtual instance. same goes for ongoing operations. how easy is it to patch the running application instances? does one need to go to each system to patch it or can it be done on one system with the patch propagating on other instances?

basically it will be all about keeping the opex as low as possible by easing the management of the cloud systems and sub systems.

more to come!

(Cloud Computing contd..): IaaS, PaaS & SaaS

the next few slides are on IaaS, PaaS & SaaS….
 
IaaS - Infrastructure as a Service

IaaS - Infrastructure as a Service

 

PaaS – Platform as a Service

PaaS

PaaS

 SaaS – Software as a Service

 

SaaS

SaaS

Cloud & Customer Intimacy

recently i came across a blog post by thomas bittman on the subject of customer intimacy ( http://blogs.gartner.com/thomas_bittman/2009/03/05/does-cloud-computing-kill-intimacy/).  it is a good post that talks about how cloud might change the need for customer -IT intimacy.

my view on the subject is as follows:-

“I think the issue of intimacy between IT and business should be treated independent of the fact whether cloud is there in the picture or not. Even in cases where enterprise IT manufactures and delivers the “services” in the traditional manner, there are some services that will require lesser degree of intimacy with the business as compared to other services.

Services that can be considered as commodity services, for example business collaboration services like email, do not require a high degree of intimacy between the enterprise IT and business even now. These services can be easily packaged based on certain characteristics (like mailbox size etc), offered based on a subscription model and can be pretty much self serving. The users can go to an Intranet, select the right package of “Service” and subscribe to the same. These can be then moved to a self-service interface.

Services which are evolving, strategic, have the potential to impact the way an enterprise goes about conducting its business or impact the enterprise end customers require the enterprise IT teams to work closely with the business functions. Such services, hence, will fall in the category of services that require a higher degree of intimacy between IT and business.

However, services that are strategic today may not be so tomorrow depending on how the service is consumed and evolves just as the case with the email service. As the service undergoes the change, so will the degree of intimacy associated with it.

I think IT needs to keep a tab on the intimacy requirements irrespective of cloud based model. these models will continue to evolve from time to time anyway. today it is cloud, tomorrow there might be something else !