KACST makes redundancy a priority

The network engineers at King Abdulaziz City of Science & Technology (KACST) spend their days ensuring two things: redundancy and scalability. The academic institute is the sole legal gateway for Saudi Arabia’s Internet and consequently has to ensure that it stays well ahead of the network and bandwidth utilisation demanded by the Kingdom’s web surfing community.

  • E-Mail
By  Greg Wilson Published  August 26, 2001

Redundancy|~||~||~|The network engineers at King Abdulaziz City of Science & Technology (KACST) spend their days ensuring two things: redundancy and scalability. The academic institute is the sole legal gateway for Saudi Arabia’s Internet and consequently has to ensure that it stays well ahead of the network and bandwidth utilisation demanded by the Kingdom’s web surfing community.

“The scalability and the redundancy are our key themes… it’s very important as we are the only Internet gateway and we don’t want to be a bottleneck to the Kingdom’s service,” says Dr. Ibrahim Al Furaih, the head of KASCT’s Internet Services Unit (ISU).

“We’re trying to make sure that we don’t reach a point where we are nearly saturated. We’re making sure that whatever environment we have is going to be sufficient long term. We don’t want to reach a point where we are running at 60 or 70% utilisation. We aim to keep very low utilisation,” explains Dr. Al Furaih.

The need for greater redundancy was brought to the fore at the tail of 2000, as several services outages — some of which were brought about by technical difficulties — hit the Kingdom. The service disruptions resulted in growing demands from Saudi’s Internet service providers for service level agreements.

KACST has sought to address both issues by delivering SLAs and signing an agreement to build a second data centre with the Kingdom’s sole network provider, Saudi Telecommunications Company (STC). “We have added another data centre as we build in more redundancy,” says Dr. Al Furaih.

The additional Jeddah-based data centre enables the ISP community to choose which centre will provide their primary connection and which one will be a backup. The two sites are linked with dual STM 1s, and within each data centre there are redundant international STM 1s. “This should provide enough capacity,” comments Dr. Al Furaih.

Running in parallel to the construction of an additional data centre has been the upgrade of KACST’s network to Gigabit Ethernet. “There have been major [upgrades] recently,” says Dr. Al Furaih. “We have upgraded all of our infrastructure to Gigabit backbone.”

To accelerate KACST’s core filtering role, the Internet custodian has deployed a mixture of Cisco and Foundry networking kit. The combination of networking kit leverages layer 4, load balancing functionality to enhance the performance of KACST’s caching farm and proxy servers. KACST also uses caching boxes from CacheFlow. “The main goal of the load balancing has been enhanced performance, we don’t want the load balancing to be a bottleneck. It’s very important… to ensure that the set up that we have here for the filtering doesn’t affect the experience of the end user. So we’re using load balancing devices to accelerate the services,” explains Dr. Al Furaih.

The combination of networking kit at both data centres has been configured in a fault-tolerant manner. The four Cisco CSS 11800 content switches split between the Riyadh and Jeddah data centres and the further four Foundry Server Iron 800 switches, split equally between each data centre, are all configured to fail-over in case of technical difficulties with the partner switch.

“We have several caches, and the load-balancing [feature] distributes the load between them,” comments Dr. Al Furaih. “This provides redundancy in case of cache failure and also provides better scalability,” he adds.

The team at ISU has been careful not to be tied into one networking vendor when constructing its network architecture. With the multi-vendor network in place, KACST is poised to offer infrastructure layer services to the Kingdom’s ISPs. “We have two Cisco switches with fail-over between them, and two Foundry switches with fail-over between them,” says Dr. Al Furaih. “Both are configured in active standby mode. What we are planning to do now is give the ISPs the choice to select one of them to connect to.”

Currently, the option to choose which device ISPs would like operating at the infrastructure level hasn’t officially been offered. There is no exact date available for when these services will be introduced to Saudi’s ISPs.

Key to ensuring the Kingdom has the maximum amount of uptime is the management layer, says Dr. Al Furaih.

The ISU’s team of network administrators have developed several tools to check the ‘serviceability’ of the network and to raise alarms if necessary. However, to deliver the single network view, ISU depends largely on HP OpenView and other vendor tools like Cisco Works.

“We have developed tools to give us a quick indication of the state of the network and monitor specific areas of the network… but the main thing that we are monitoring is network capacity and we manage this by our own tools,” explains Dr. Al Furaih.
||**||

Add a Comment

Your display name This field is mandatory

Your e-mail address This field is mandatory (Your e-mail address won't be published)

Security code