Advertisment

Servers Consolidate and Conquer

author-image
DQI Bureau
New Update

It is better to practice than to preach and we decided to do the same. Rather

than talk about the server consolidation benefits of other companies, we decided

to look in-house–at Cyber Media, the publishers of Dataquest, and study the

possible benefits of server consolidation.

Advertisment

The Cyber House story



At Cyber House, the headquarters of Cyber Media, there are four HP E-class

servers, one HP LH-class and four assembled operating multiple platforms:

NetWare for File and print, NT for Oracle database, mail, intranet and backups

and Linux as the web proxy. The processors in the servers range from Pentium 75

to PIII 866. The storage across all servers is 96 GB. Apart from the

infrastructure, Cyber Media is planning to deploy a CRM package, enhance the

current circulation software and make it web-enabled and implement a workflow

application.

When to Consolidate
Is your company’s server infrastructure ready to be

consolidated? If yes, when is the right time to do so?
  • Whenever there is talk of capacity expansion, you think of new

    servers.

  • It is the same with scalability. If you need

    scalability, you need

    bigger servers.

  • In the outsourcing age, your IT staff budget is equivalent to the

    sales and marketing department staff budget.

  • The users in your organization are very IT savvy and buy their own

    hardware but managing it is your headache.

  • You are the proud owner of servers with different

    OSs.

  • Utilization rates for a majority of the servers are in single

    digits.

  • At the end of the month, you have to make a physical verification

    of all your servers.

  • Server maintenance is eroding the bottom line.

  • TCO is becoming an important topic for discussion with the

    management, apart from vendors

  • Non tech savvy users start talking about SLAs

Given this background, what happens if we go in for server consolidation,

assuming of course, that we converge to a common OS platform?

Advertisment

We decided to look at two HP LH3000s for consolidating the server

infrastructure. The specs —1 Ghz two CPU processors, 1 GB memory. The machine

comes with hot-swappable/hot-pluggable components, including disk drives, power

supplies, fans, and hot-plug PCI slots.

We also put in a 36*3 GB raid array. The cost is about Rs 6 lakh roughly,

without factoring in discounts available today.

The benefits: Firstly, Cyber Media would reduce the multiple points of

failure from the nine servers to a more reliable and robust single system with

very high redundancy. Since it will be a single system, more resources including

power, can be allocated to it. The target now, is to provide a single system as

against the nine systems that existed earlier, so that it provides multi-level

power back-ups easily. This would also mean saving on the cost of power

consumption. Moreover, Cyber Media can provide additional online backups with

the RAID 5 arrays.

Advertisment

Also, since Cyber Media has a heterogeneous environment, this

will also result in savings in terms of cost of redundant licenses. Of course,

these systems will be high scalable and reliable, thanks to a host of features

like hot swaps and other high availability features. The server with intelligent

manageability tools will help in lowering maintenance and running costs, lower

expertise and the man-hours needed. Given the high availability and reliability,

downtime can be significantly reduced and better service levels can be

guaranteed.

Also, since we are planning to deploy other applications, we need to factor

in the opportunity costs missed as a result of not undertaking server

consolidation.

As the volume of transactions is expected to shoot up, the investment in the

new servers will be protected and the actual payback period will be much lesser

if these are also factored in.

Advertisment

Have we left out the current AMC cost? Well, the system will come with a

three-year warranty (so no AMC cost). So, even in a medium sized setup like

ours, server consolidation gives direct payoffs within a short time.

The heart of your network



Servers are the most important component of IT infrastructure. They form the

heart of the network and unlike their cousins, the PCs, often live for 10 years

or more. And how important are servers? CIOs don’t have the luxury of taking

servers offline for repairs. So server purchases will continue to drive the CIO’s

budget.

While servers are going to be an important part of this budget, this time

round they will be driven by the need to buy bigger boxes aiming to consolidate

their operations from the distributed computing model to the centralized model.

Increasingly organizations are realizing that while distributed models work,

they also increase the total cost of ownership.

Advertisment

To cut down the same, companies are going back to the mainframe days…

Things are best handled if resources are located in a single, central place.

For whom?



Server consolidation gives the organization huge benefits (See box) but it

is imperative to decide if your company fits the bill for the process. It’s

ideal for organizations with ERP implementations, large telcos, banks, large

organization, government, and



educational campuses. However, even mid-sized companies that have multiple
branches and 250-plus users can benefit from the advantages of server

consolidation.

Benefits of Consolidation
Higher the level of consolidation, higher the gains. Here’s

how you can save up:
  • Cost of staffing needs

  • Cost of upgrades.

  • Reduction in administration costs.

  • Reduction in the costs of delivering service.

  • Scalability, allowing for prompt reaction to changing business

    needs

  • Reduction in systems management costs including maintenance as well

    as improved systems management.

  • Improved availability of service (fewer or no outages, and shorter

    outages should they occur)

  • Back-up and security issues.

  • Better utilization of floor space.

  • Lower power bills.

Advertisment

Alternatively companies with six to seven servers, like Cyber Media, with

significant AMC costs, can look at server consolidation within the same

premises. There is no thumb rule on the bare specifications for server

consolidation, but this will certainly need to be weighed in terms of the cost

and the benefits–direct and indirect.

Types of server consolidation



If you are convinced about the merits of server consolidation, it is time to

look at the type of consolidation that would suit your business objective.

Broadly, consolidation falls under four categories. Organizations will have to

assess and check which model will fit their requirements best.

  • The first one is logical consolidation, aimed at reducing the number of

    control points in the environment to a single administrative stream. So,

    while the servers remain dispersed, local operations are reduced or

    eliminated, and management functions such as backup, restore, recovery,

    maintenance, and user support are performed remotely.
  • However, physical consolidation involves consolidating a

    single site. The user replaces many smaller servers in the same architecture

    with fewer, more powerful servers, usually relocated to a single server

    site.

  • Application consolidation involves bringing all the

    application in a central place and reducing the need for multiple servers

    for the applications across the various offices of the organization.

  • Workload consolidation on the other hand involves

    reducing not only the number of servers, but elements such as tapes, disks,

    network devices and connections, software, operating systems, peripherals,

    as well as the number of processes and procedures.

Advertisment

With fewer hardware and software standards to manage, IT

departments can more easily move systems, applications, and peripherals.

Options available



This is the easy part. In terms of PC server space, apart from the

international vendors like IBM, HP and Compaq, many Indian players are also

active in the field viz. HCL Infosystems, Zenith, CMS et al. However, in the

RISC space, international vendors have a complete monopoly, with Indian players

like Wipro only distributing and integrating them. Given the current slowdown in

the industry, pricing issues are in favor of CIOs. Even with slashed budgets,

CIOs have still gone ahead with projects they had planned in the beginning of

the year, thanks to falling prices. Also, vendors who want to gain entry into

organizations are offering cutthroat pricing–to the point of ‘stripping’

(60-80% discounts) to win the account.

No thumb rule for server selection



Is there any thumb rule for server selection? Today, the first preference in

any new purchase is the buyer’s relationship with existing vendors. Depending

on the comfort level here, CIOs decide to stick with the vendor or pick a new

one. However, if you are looking for servers from other vendors, there are

several parameters to look for before zeroing in on your choice of server.

First, the much-touted benchmarks are not high on the list. Since each vendor

will produce specific benchmarks where his product tops the chart and the

results do not portray real-life situations, it should hardly have an impact on

the buying decision. Talking of real-life situations, a better test is to run

the server continuously for 72 hours with specific applications. If it works

fine for the given time-frame, it will work for the next couple of years without

many glitches.

Best Practices
  • Support: Check

    with your peers on their experiences with

    vendors regarding support and service issues. Insist on a comprehensive

    service level agreement (SLA).

  • Scalability: Whether buying for consolidation or not, do keep a roadmap

    of possible future applications and have the scalability built in

    accordingly.

  • Avoid multiple

    platforms
    : If you have multiple platforms floating in

    your organization, integration and management of various applications can

    become your worst nightmare.

  • High availability

    features
    : Look for servers with ‘hot-swap’

    components, as they are ideal for mission-critical operations, where you

    cannot afford any kind of an interruption. These allow you to swap

    components while keeping all the operations running and thus helps avert

    system failure.

  • Performance: This is easy, thanks to the tons of benchmarks available.

    For the processor, have a look at the SPEC, and the application level is a

    good indicator of system performance. There are benchmarks like TPC-C, TPC-H,

    SPECSFS and others like SAP ATO and SD.

  • Investment

    protection
    : Before you make the final decision, also check

    if the vendor is guaranteed future investment protection. If you are

    looking at the high-end RISC servers, hardware partitioning and binary

    compatibility features ensure that upgrading the system would not cost as

    much as replacing it. With Intel servers, this may not be the case.

If you want to upgrade two years down the line, you can buy older processors

and scale up the system. The only problem with this approach is that older

processors will be more expensive than the current generation of processors

because of lack of economies of scale. You will pay more and probably get

one-third the performance of current-generation processors.

In most RISC servers, the facility of hardware partitioning takes care of the

same. With Intel servers, options are



limited on this front. However, the advantage of going with Intel has been the
constant fall in prices and more powerful configuration.

Where to?



Servers are a must. No second thoughts on that. Also, thanks to the

distributed computing model and low-prized Intel servers on the Microsoft

platform, India–much like the West–is witnessing server proliferation and

sprawl (commonly referred to as the "mushroom effect"). With

medium-sized companies growing and IT driving this growth, this is expected to

continue unabated. However, most organizations may still spend or end up

overspending on server infrastructure and support. But does server consolidation

really save organizations money and time, while improving performance and

increasing IT flexibility?

This is not what the vendors said when driving the distributed server-based

computing model for the last decade. And more importantly, distributed

processing works well. Why stop now? Is it just another hype being created by

vendors to sell bigger systems to the decentralized organizations? No doubt, it

is bringing in new businesses for vendors, but it is also a way out of the

systems-management chaos. Studies show that the acquisition cost of a server

represents only 15-20% of the total cost of ownership.

Oops! Did we forget to mention price? But then, you shouldn’t really be

thinking money when you talk about the heart of your network. Anyway, discounts

today make servers cheaper than ever.

Not just for the big guys



And server consolidation is not restricted only to Fortune 500 companies.

Even smaller companies can reduce TCO by server consolidation. It is true that

the bigger the scope for consolidation, the better the return on investment and

the quicker the payback period. But there is hope even for medium-sized

companies with 8-10 servers across the country. According to the Gartner Group,

consolidating six small servers into a pair of larger machines provides

substantial TCO savings of 35-40%, primarily in internal support costs. However,

server consolidation represents the only cost-effective way that you can meet

level-of-service requirements associated with applications needing scalability

and resource optimization.

If you are planning on buying servers, keep in mind that other organizations

have started looking seriously at server consolidation as an important strategy.

If you have started witnessing systems-management related problems, its time to

explore server consolidation for your organization. If you have still not

reached that stage, it good to plan for the future and roll out the server

roadmap with server consolidation topping the priority list.

Team DQ

How to do it?

Feasibility

Assessment: What tells you if your decision is the right one?

  • Understand management concerns about

    service level delivery (uneven data integrity and security, outages of

    significant duration or difficulty of developing new cross

    applications), availability and cost containment.
  • Scope of effort based on an analysis

    of environmental and system complexity, size, organizational

    structure, configuration including server, network and database

    structure and operational complexity.
  • Identification of consolidation

    opportunities including a review of options for various applications

    and workloads.

Cost-benefit

analysis
:

The most important component of any analysis. Will the benefits outweigh

the cost? Are the benefits tangible and can they be quantified, or are

they intangible? For a rider, the assessment of benefits due to lower

complexities and cost needs to be done over a 2-3 year time-frame against

likely costs. Also, the company needs to look at the opportunity cost lost

if server consolidation is not undertaken. Benefits like reduction in

staffing needs, and cost of upgrades need to be factored in.

Vendor

choice
: Before

calling in the vendor, have a look at application requirements, database

requirements and architecture of the platform on which the system is

based. To do so, gather data from each of the applications and servers,

which are candidates for consolidation, including application and system

design requirements, as well as service level requirements. On the

services offering half, have the performance measurement data analyzed for

each of the systems and workloads–the ideal candidates for

consolidation.

Advertisment