Real-time infrastructure is the ultimate objective: being able to allocate and deallocate resources automatically and in real time. The problem currently is that the infrastructure is only seen as plumbing.
Server evolution
Huge amounts of money are spent but there is also a tendency to commoditization. From mainframes (still alive and well) to blades, it is getting more complex, with virtualization, compliance & regulation, management…
Affecting the landscape are cloud computing, software appliances…
Blades are still proprietary but it is becoming componentized (although they remain proprietary). Blades will soon touch everything: server, storage, switch, memory…
Blades are the highest growing area in the server space but it has a power problem (a cabinet will require 25 to 30kW of power in a few years).
Client evolution
This is a battle digital natives & immigrants: the former will want to bring their "toys" at work. It translates into a battle between losing and loosening control.
There are also emerging "devices" (non-enterprise owned, mobile, bubbles - virtual machines for different environments -, streamed applications).
IT must still provide the basics (minimal management, security).
Green IT (power & cooling)
Green IT is and is only about energy management, for the short term, essentially because of the energy cost issue.
There are also problems getting the power needed for new deployments. When planning for the future, you've got to make the assumption that the energy consumption will go up (x2 in 4 to 6 years).
Virtualization
This is the number 1 subject for Gartner clients. Rakesh presents a comparison table between VMWare, Citrix, VirtualIron, Oracle, Sun, Microsoft on major criteria (VMWare and Microsoft are considered the most important to watch now).
This is a huge market with a lot of potential. Virtualization will go from 8% (of x86 servers) today to 50-60% in 5 or 6 years.
There will be a high level of heterogeneity because of applications delivered "virtualized".
The "V-Hive" shows the diversity of the ecosystem. Not one vendor covers every domain. Make sure your ROI is short, many vendors will disappear.
Data Center networking
A new ethernet-based standard is emerging (I must admit I probably missed a point here), along with new devices managing security, compression…
Storage
Thin provisioning helps improve utilization (up to 65-80%) and automating storage capacity management. The storage virtualization level remains low today.
Another subject is about optimizing capacity and reducing the volume, using deduplication, single instance store, compression.
The main issues for storage are: energy consumption (but still, remember the real energy issue is not on the storage side), storage costs, space, data availability, performance.
Management repositories
CMDB is critical but the way it is used is often not right. It allows to make good impact analysis, provides a common asset database also showing relationships between the assets.
You have to make sure that your data sources can be trusted and that the CMDB is accompanied by change management processes to keep the DB up to date.
The PMDB (performance management database) will become mainstream by 2012. This will help make better decisions.
Automation
There are a new set of technologies for improving the business alignment (e.g. on SLAs), predictability, supporting best practices, saving time, mitigating risk. These tools are federating the whole picture of change management.
It is already used for disaster recovery.
It is worth assessing in large data centers.
SaaS & cloud computing
There is not much to learn here, compared to yesterday's talk except that the associated risks are pointed out (the data being externalized, the issue with service level agreements…).
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.