News

HMS Is Facing a Deficit. Under Trump, Some Fear It May Get Worse.

News

Cambridge Police Respond to Three Armed Robberies Over Holiday Weekend

News

What’s Next for Harvard’s Legacy of Slavery Initiative?

News

MassDOT Adds Unpopular Train Layover to Allston I-90 Project in Sudden Reversal

News

Denied Winter Campus Housing, International Students Scramble to Find Alternative Options

HASCS Revamps Network

TechTalk

By Baratunde R. Thurston

"My e-mail is broken!"

How many times did you find yourself uttering this last year? Let's face it: there were times when a letter delivered by Santa's Little Helper would arrive at the Quad faster than one sent via pine.

With over 15,000 active accounts, the old system had been stretched to capacity, and did not handle failures well.

There were four main machines that handled everything: you logged into fas, while husc handled your e-mail, a machine called the nab held your home directory and scunix4 served the workstations known as scwsl-44.

These servers were really big, and when one of them halted, the usual result was that everything would come to a crawl or stop altogether.

But times have changed. The big boxes are out, and clusters are in, with the advantage being redundancy and scalability (room for growth).

The two new file servers (corel and core2) replace the nab and each constantly makes sure the other is running smoothly. If corel notices that core2 is failing, it will take on all of core2's load until core2 is back up. Don't you wish you had friends like this?

The general login process has changed as well. When you telnet to fas.harvard.edu, you are now sent randomly to one of six new machines dubbed login 1-6.

In the old days, if fas went down, everyone felt it and was shut out. Now, if loginl crashes, all incoming telnet requests are immediately routed away from login 1, and it is removed from the cluster for servicing. And more login boxes can be added to increase capacity.

Similarly, there is a cluster of three Web servers, four mail servers and four course servers.

The latter are for students who need to use UNIX systems for their course. They work should login to ice.fas.harvard.edu ("ice" stands for instructional computing environment) and are randomly sent to ice 1-4.

In addition, there are 33 new workstations (dubbed wsl-33) to replace the scws machines.

Beware, though--these machines are no longer the fastest offered for course work. Unless you need the large screen, you are better off programming and compiling one ice.

--Baratunde R. Thurston '99 is the Claverly Hall User Assistant for HASCS, editor-in-chief of Computers@Harvard, published by the Harvard Computer Society and a Crimson editor. No, he does not have a nickname.

Want to keep up with breaking news? Subscribe to our email newsletter.

Tags