Subscribe

0

Features

Servers

author-image
DQI Bureau
New Update

Plan for excess capacities.

The needs of your organization are expanding at a rapid pace. Capacity
planning for future workloads (at least for serveral years) is therefore
important. Application servers can cluster a host of servers and treat them as a
single pool of resources, in which when one server goes down, the other server
takes up the extra load, thus ensuring availability of services throughout the
day. This reliable production environment is especially important for the
functions that demand round-the-clock availability, such as online services and
transactions.

WATCHLIST

Standardize

Even with only a few servers, a standard server spec gives you interoperatibility and
interchangability, easier support, common standby spares (such as power supplies), and common upgrades. At the very least, try for a common vendor

Negotiate

With servers, this is not about price. It’s about support and uptime. The key things to negotiate are the nature and length of the warranty, and the service level agreement. This many include the vendor agreeing to stock key spares on your premises

Outsource

In facilities management, the servers are a key area. Can your vendor or support service guarantee 99% uptime? How about 99.9%? (and even that’s nine hours of downtime!) Outsourcing helps: most small to medium enterprises will not be able to demand such service levels from internal IS departments

Consolidate

Bring your apps together on fewer servers. “Spreading” your points of failure is old-fashioned now. A single, high quality high-power server with a RAID-5 array is better and more reliable than three smaller servers adding up to the same price

Backup

Power try using smaller secondary UPSs closer to the servers, in addition to your main enterprise UPS, to insulate the server from primary UPS failure. The local UPS should have an auto bypass, and a serial link to help the server shut down gracefully. 

Data backup is crucial–even if users back up their own data, server structures, NDS and ADS tree, et al, must be periodically backed up

Seek high-availability features

Look for servers with ‘hot-swap’ components, as they are ideal for
mission critical operations. It allows you to swap components while keeping all
the operations running and thus helps avert failure. IBM, Sun, HP, and
Compaq/Digital (all Unix) currently provide high availability through clusters.
These provide you with mainframe-type availability, with hot swappable
components, including power supplies, processor, memory and fiber channel IO
cards.

Useful Tips

  • Have the operating system pre-installed. Look for pre-configured
    small business server packages with additional server software
    thrown in. While these are more useful for smaller businesses, a
    medium-sized enterprise could use these as workgroup servers.
  • If your principle need is to share files and printers don’t
    bother with Multi-Processor servers. These only come into their own
    when you intend running an application such as a Web server,
    database engine or E-mail server.
  • Go for redundancy with multiple power supplies and extra fans if
    available.
  • A fast network adapter is essential. Opt for a fast Ethernet card
    that can support 100Mbit/sec as well as 10Mbit/sec connections.
    Multiple NICs can be installed to provide useful protection against
    hardware failure, as well as boosting server bandwidth.
  • Basic monitoring and management is a must, on top of which look
    for remote management facilities and Simple Network Management
    Protocol (SNMP) support.

How much power?

How much power?

Features

Low-end Server
  • Single purpose machine.
  • Limited upgrades available.
  • Designed for remote sites, not always in clean environments.
  • Remote management key criteria in design and construction.
  • Local disk, tape units, internal to the server (rack mount).
  • Due to size and low cost, multiple points of failure is tolerable.
  • Servers deployed easily/repeatedly for business convenience.
  • Outages tolerated from 8-16 hours, spares and field support designed accordingly.

Mid-range Server

  • One large or 2-3 small functions can be consolidated.
  • Upgrade potential, up to 100% (CPU, memory).
  • Server is essential to business, hence controlled environment.
  • Tape still likely to be internal.
  • Some redundancy with power supply, RAID, network interface. Outages not to exceed 4-8 hours

High-end Server/Enterprise

  • Multiple CPUs configured to handle multiple workloads.
  • Job schedule and control software required.
  • Upgrades and clustering to provide 300-400% scalability.
  • Server in data center with critical environmental and security control.
  • DASD will be external, sharable, extensible and highly redundant.
  • On-site 24x7 technical support staff, operations and systems programmers.
  • Redundancy in CPU, Memory, channels, power etc. to provide

    over 99.90% availability; Unscheduled down-time not tolerated above 20-30 minutes
Enterprise server is the same as a large high-end server except that processing, storage, operation systems and applications intended to be non-stop in nature. 99.8% availability expected. Tight physical security required.

Depending on the number of employees in your company and the kind of
functionality they need, you can choose between light or low-end, medium,
enterprise or high-end servers. For instance, while a light server might be
sufficient for messaging or mail applications among 500 people, for database or
data mining, such a server can only support about 50 users.

Look for open platforms

Integration and management of all kind of applications running across the
organization on varied platforms could be challenging for any enterprise. You
need the ability to integrate existing backend databases, transaction monitors,
message queues, directory services etc. as well as adhere to popular standards.
An open architecture system is extremely useful especially in the case of large
enterprises that need to work on multiple standards. The applications and
components developed on any standard can then be easily integrated into the main
system. In the case of Web servers, for instance, Java technology-based
application servers have already proved their compatibility across multiple
hardware/OS configurations, though it is a general perception that they are most
stable under Unix environments.

Server consolidation for minimizing costs

Server consolidation — bringing together applications, databases and
services onto fewer, highly reliable severs — is one of the hottest trends
today. While IT managers still continue to work on the principles of distributed
computing, they are also returning to the notion of centralized design and
management. Consolidation enables you to economize on resources, reduce the need
for floor space, increase control and improve efficiency at a lower TCO.

Monitor your critical transactions

Most enterprise systems have to run a series of database interactions or
transactions and some of them could be extremely sensitive or critical. Keeping
a control over them involves complicated mechanisms. A mainframe environment
would have has a separate box called the Transaction Monitor to perform this
function and modern day application servers have built-in transaction monitors
or work in tandem with other transaction monitors that are already up in the
enterprise. When the transactions happen between multiple application servers
spread across multiple databases, monitoring and exercising control becomes
important. Although it is a tedious process, the success of an application
server depends on how well it can protect the transactions from different levels
of failure.

Prepare for E-business

E-business pressures have resulted in various changes in systems such as the
removal of batch windows and reduction in time available for system maintenance,
backup, and recovery. As a result, many organizations are moving towards more
open infrastructures such as SAN (Storage Area Networks) or NAS (Network
Application Systems). E-business is about integrating servers with the Internet
and critical applications with data from suppliers and end-users. Whether you
are a dot-com startup or a brick-and-mortar company on its way to e-enabling its
business, you need tools to manage your e-business plans. You need to devise
ways to manage end-to-end risks and costs, while you incorporate the new
emerging applications on to your online setup. The only way to get this level of
integration is to build an e-business infrastructure on an open,
standards-based, scalable software platform, which allows flexibility.

A DQ report