|
There
are many benefits of having your entire IT infrastructure
as a virtual environment. Ooi Sze Kai, Brand Executive,
Storage Systems Group, Asean/SA elaborates on IBM's
efforts in this direction with their on-demand strategy
for storage virtualization. by Anil Patrick R
What is IBM's On-Demand Vision?
According to IBM's On Demand
vision, an enterprise should achieve a total integration.
The integration should be end-to-end of its company
processes from the supplier to the customer. If the
customer has established this capability, then the benefits
are many. He/she is now able to react with great flexibility
and speed to the evolving changes in the market place.
For example, in the manufacturing
sector you have just in time. This is the ability to
bring in just enough supplies to meet exactly what the
customer is buying. Similarly, a business using on-demand
is one that has totally integrated its entire processes.
So the change on this side flows right down to the entire
process. With this he/she has a real competitive edge
over the competition and high efficiency. IBM wants
to provide products and solutions to achieve this, and
services to help the customer during his process.
What are the requirements for an
enterprise to grow to the on-demand environment?
In order for businesses to grow
to the on-demand environment, there should be some provisions
created in this environment or infrastructure.
First of all, there must be
a tight integration between processes and data infrastructure.
Processes must be end-to-end. Second, the customer must
establish a completely open infrastructure. It should
be possible to put disparate stuff together and integrate
them together. This is possible only if standards are
open. The third area is to put intelligence to make
a virtual environment. It is necessary to transform
the entire setup to appear like a cloud, so that it
is virtualized. This is similar to the Internet world.
Anybody plugging into that cloud knows for a certain
fact that he will be able to communicate irrespective
of the hardware he uses.
An on-demand environment needs
to have autonomy. Autonomy means making the cloud so
intelligent that it can sense and respond to the changing
demands of the user. When you have such a big cloud,
it has to manage itself. In storage, once the cloud
is created, it must be able to sense, respond, protect,
and tune itself. It must know when a user comes in,
what resources to give, protect, ensure service levels
for the user/server/client that is plugging in, etc.
These four attributes are true
whether it is establishing an on-demand environment
for servers, storage,
or network.
What are the major components of
storage on-demand, that you have just announced?
The first announcement is a
virtualization capability called the IBM TotalStorage
Virtualization family. The products are the SAN volume
controller and the SAN integration server.
The second area is the announcement
of a common file system. The product is announced as
the IBM TotalStorage SAN file system.
The third product is a common
management system that allows open management of devices.
We have announced a common management tool so that any
manager can manage any resources as long as Storage
Management Initiative Specification (SMIS) standards
are adopted. They can be managed by a common agent.
What can the Volume Controller
and Integration Server provide?
The SAN Volume Controller is
an appliance server to provide virtualization capability.
The second product is an integrated version called SAN
Integration Server. Inside the server, there is a combination
of volume controller put by the engine, along with a
switch and SAN switches. If block aggregation can be
virtualized, the capability to present a physical disk
as a logical volume to the server is possible. This
is the first step, which is done by the virtualization
engine.
Everything can be done at the
top itself by moving intelligence to the network. The
SAN Volume Controller does standardization at the top
itself, which is not possible with traditional methods.
This is enough to present logical volume to the top,
and attach to different device or server types. It is
also enough to do advanced functions like mirroring,
copying, snapshots, etc.
What are the advantages of putting
the intelligence at the network level instead of at
the disk level?
The biggest advantage is that
intelligent disks can be done away with. More flexibility
comes in. It is possible to transform the meshed environment
into a cloud. The virtualization engine inside the cloud
provides all these capabilities. All the servers on
top can map into the logical volume below by working
through the cloud.
Another advantage is improved
administrator productivity. Using the cloud, all that
the administrator has to do is manage the engine. Physical
and transparent management from the application server
is possible. All the application servers talk to the
engine and are presented with a logical volume. The
importance of this is that if a server needs more capacity,
it can be done on the fly. In this environment, the
virtualization engine performs all the background work.
The application does not even need to know since it
still sees a logical view and the connection is not
lost. This literally translates to higher availability.
The second advantage is much
higher efficiency and utilization. In the earlier days
of SAN, even though organizations could share data,
what actually happened was a concept called 'Zoning.'
Lots of free space had to be
provided in case the application server went down. The
extra space cannot be shared since it is related to
the direct line of connection. This leads to a situation
where the SAN utilization of the physical device is
very low. Now in the virtualized environment, the actual
relationship is only between a server and a logical
volume. There is no need to partition free space related
to one device since free space can be shared by all
servers. Therefore, we find that typically in moving
from a SAN into a virtual storage environment, you can
actually increase utilization in a major way.
It will also enable a common
platform for copy services like mirroring. Organizations
with large availability need to have data exactly the
same way at their disaster recovery (DR) center. This
means each disk has to mirror another copy elsewhere.
Every disk vendor has their own mirroring formats with
different licenses, software, etc.
By going in for virtualization,
since the engine is managing the relationships, it just
has to talk to the engine on the other side. The advantage
of this is that I can mirror an expensive device into
an inexpensive device, since the DR site will not need
cutting edge devices. It also means lesser licenses
and lesser manpower to manage them.
What benefits can the TotalStorage
SAN file system provide?
There are different file systems
out there leading to multiple copies. The common file
system addresses the capability to share different kinds
of files.
By doing this it is possible
to really extend the value of virtualization right down
to the file level. This is most important because the
system administrator gets the capability of more volume
no matter how many platforms he runs. This is where
the common file system can really help. It is basically
a metadata server to keep information about files. It
will then guide you to the virtualization engine, which
will find the file for the user.
What happens is that there is
an agent that is installed in each of these different
servers for the common file system. Whenever an OS or
an application wants to open a file, the agent will
direct him to talk to the metadata server. The metadata
server will translate it and tell the OS from then on
how to get along with the file and what to do. This
means that the storage administrator can now define
policies at a file level. From a single point he can
now control the policies for all the files in the enterprise.
This is very useful in terms of aspects like file access,
security, backups, heterogeneous file sharing, etc.
Anil Patrick can be reached at anilpatrick@networkmagazineindia.com
|