net.work

The Way Business Is Moving

net.work published by
Issue Date: May 2002

Network availability and performance: a QA with Controlware's Wiebe de Vries

May 2002
Lucy Hunt and Wiebe de Vries, Controlware Communication Services

[Q] Network reliability and performance are crucial for operating mission-critical business applications. What options are available to businesses to improve network reliability and performance?
For mission-critical applications a number of options are available, depending on the requirements of the network. If uptime is an issue business have ample choice of UPSs, mirrored servers, replication etc. as long as it is in the same network. As soon as we think of 'remotely connected networks' the issue becomes a different one since then carriers become involved.
A lot of companies are starting to run VPNs to simulate a LAN even though the connections could be very remote. Users depend on information being available, and in the case of some applications the time-outs between request and response become critical. Not getting a reply in time can cause an application to crash, which means the databases in that application could become corrupted, which means high cost for the organisation involved since users will be sitting still waiting for the application to be restored, and IS people will go into panic mode and drop all structural activities to resolve the issue.
Considering this, the investment in equipment that provides 'continued connection' on the network, even though it may seem high initially, can be calculated along the same ROI guidelines that are adopted to other strategic investments.
Disaster recovery or back-up equipment that ensures continued connections is available for a multitude of infra-structure types, be it ISDN, fibre or fixed copper connections. Which one is chosen depends on availability, cost of renting a connection that (if all is well) will never be used, and the cost of the connection should it be used for a back-up application.
There are also scenarios that can be adapted to the way the recovery is done. Either the connection between the location and the nearest PoP is restored, or the connection is completely made point-to-point (or end-to-end) meaning that the existing network is eliminated from the equation until it has been verified as being available again at a service level that is required by the application or the network.
In the case of back-up of existing leased lines, the cost of the leased lines is also an important factor. Getting a second leased line as back-up means involving a second carrier in the network, which creates management issues for the business involved. ISDN is widely available and has also proven itself as a reliable network. The cost of the subscription is not exorbitantly high, and the cost of transferring data can be offset against the cost of not having a connection at all in case the leased line fails.
The approach to 'dial around the cloud', ie establish a completely controlled connection is also advisable since it is highly likely that the loss of leased line connection occurs in the PoP and therefore trying to establish a back-up connection to that same PoP might succeed but the connection from the PoP to the backbone will be interrupted, thereby reducing any chance of a successful back-up connection. It is for this reason that Controlware has adopted the 'dial around the cloud' approach.
[Q] Carriers are committed to delivering the highest level of service and respond to corporate demands for network reliability, scalability, and flexibility. How do you think this will translate into business and service opportunities for South African businesses?
In the continued quest for providing quality of service (or uptime) and the upcoming liberalisation of carrier markets the deciding factor for any company offering connections will be how well it handles loss of connection for its customers. It could be an offering by the carrier itself, or by ISPs establishing themselves as incumbent carriers/ASPs.
In all instances the customer demands Service Level Agreements specifying a guaranteed availability of the WAN. Availability can only be guaranteed if the provider offering the SLA has control over the way downtime is addressed, or better yet eliminated. Either the provider specifies that there must be a back-up scenario, or the provider could decide not to trust this important part of the offering to any other business than his/her own and make sure that they control the way line failures are addressed.
Meeting this SLA is a requirement to get the business, and offering better SLAs than the next competitor is a differentiator in the business offering.
In many cases the SLA is becoming a part of the complete 'managed service' offering, where the end-customer will have equipment installed for LAN and WAN purposes for which maintenance and management is completely outsourced to a provider of Managed Services. Any provider of Managed Services will want to make money on the offering that they do. Being able to handle a lot of things remotely is one certain way of saving costs in case of an incident, since they can verify the cause of the incident before going on-site, and when on-site due to pre-investigation are able to solve the issue quickly. Having equipment in place that is intelligent enough to establish back-up connections based on a multitude of criteria supports the quest for lower expenditure and higher return of investment, and the possibility to offer Managed Service at relatively low costs.
[Q] Corporations have to be able to keep mission-critical applications running in the event of network failure. Disasters happen. How does ISDN Backup and Access help in the event of catastrophic network failure? If at all?
In the event of catastrophic network failure it is questionable whether anything would help. However, it depends on the definition of catastrophic. For a company that depends on network connections in order to do business (on-line sales systems, ATM machines etc) not having a connection is catastrophic already.
[Q] Finally, to what extent do you think business availability, business continuity and disaster recovery have moved from being back room IT issues, to top of mind executive issues? (Should they be top of mind executive issues? Why should we care?)
Business continuity has become an important issue since the happenings of 11 September. For Controlware it has been an issue since 1989 when the first units for this specific application was released. The shift in perception is seen clearly from the fact that, where in the beginning of the product line the customers were exclusively carriers (using it for increased SLA), now there are more and more end-customers that want to address the issue. It has become that way when the economy started 'normalising' in 2001 and subsequently incumbent carriers started declaring bankruptcy or insolvency. It pushed the message home that IT managers and executives need to be sure that the issues have been addressed properly instead of assuming that 'someone will take care of it'.


Others who read this also read these articles

Search Site





Subscribe

Previous Issues