Wherein we write down some stuff that we know.

Archive for the ‘Portal’ Category

Monday Isn’t the Start of Busy

Friday, August 24th, 2007

The fall term officially starts on Monday, but our portal servers will be very, very busy Sunday night as student and faculty get online to check their schedules and classes. Last year at this time we had well over 1,000 concurrent users in the portal for most of the night, as you can see below.

Sunday was crazy

The first Monday is even crazier, as one would expect. Last year we even had to take one portal server down for maintenance. Our current setup allows us to do this without much fuss, as we just roll in a hot stand-by server.

Monday was crazy too

Portal Upgrade Complete

Tuesday, August 21st, 2007

As the summer winds down, WEBD has successfully completed the uPortal upgrade that we’ve been diligently working on for the past month. In addition to moving to the uPortal 2.6 framework, and DLM based layouts, we’ve improved the presentation layer quite a bit.

The most exciting element of the upgrade is the move away from table based layouts thanks to the Yahoo! Grids framework. Thanks to some clever XSLT hackery we are now able to support 1-3 column layouts. Plus, we reimplemented our tabs using the YUI Tabs mark-up and styles to take advantage of the browser testing Yahoo! has already done and standardize on something.

We also made a slight change to the location of the navigation bar by moving it above the banner graphic to a location more inline with where major websites (Yahoo, Google, etc.) place the ” YOURNAME Login/Logout” links.

Not content with our footer, we jazzed up the mark-up with some Microformat goodness and tossed in a Campus Search box; just in-case someone has a need to search for something not in the portal.

The finishing touch of the upgrade was the migration away from Verdana as the main sans-serif font, to Helvetica Neue for Mac users and Arial for Windows users. The migration to Helvetica was a small homage to the Swiss school of design who brought us such tools as grid frameworks, which we have employed in the portal layout.

AlcoholEdu, The Search for

Monday, August 13th, 2007
Note: For those students looking for Alcohol Edu, please login to the portal and go to the Records, Registration, & Finances tab. There will be a link to Alcohol Edu on the right. The reason that it is in the portal is that the portal provides a single sign-on to the service so that you don’t have to fill in most of your information again.

Recently ‘AlcoholEdu’ and ‘Alcohol Edu’ came up in our popular searches. It’s not surprising as most of the communication for this program is via paper that is mailed to students. I know that when I was a student, paper mail got lost after I got it. A lot. I imagine things haven’t changed much since then.

Without much information about the program on our web site, searches are turning up mostly irrelevant information. The first thing we recommend is to create a page with all the information that people would need about this program (times, dates, links, etc.) and link it prominently from the department main page. The title of the created page should also be specific (“AlcoholEdu Information”) and make good use of header tags (h1) to convey to the search engines the purpose of those pages.

If people are searching for your information that doesn’t necessarily mean that you or your organization have failed in your organizational scheme. The age of search is upon us. I suspect the number of people that try to find things via navigation is dwindling, while the number who first try search are growing. There is a good chance that people aren’t even going to your site before they do a search. You need to make sure your site works for the searchers.

SPOF/SPOC Redux

Thursday, August 2nd, 2007

This came up in a discussion between Scott and I today and I realized that it would be good to provide the context in which the statement was uttered.

When that sentence first spewed forth from my frontal lobe, I was attempting to convey the advantages (at least in our heterogeneous university IT environment; unknown for others) to bottlenecking users through a single point for access to enterprise-wide applications. By looking at what might be referred to as a single point of failure instead as a single point of communication, the notion of having a single container in which to place context-specific messages about these enterprise apps is made visible.

Really, this is just a re-wording of the fundamental purpose of a campus web portal, but after our horrific experience with Campus Pipeline, significant resistance had arisen to the single gateway concept. (Some out-of-context legacy IT mumbo-jumbo about avoiding SPOFs that might have been true in the 90s fed this resistance as well.) People responsible for these other enterprise apps wanted to make sure that the non-portal avenues to their apps were widely publicized. They were questioning the reliability of the new Portal. Fortunately, with a mixture of CAS, some load-balancing content switches, and (most importantly) top-end admin/developer and network talent, those dragons were slain.

Portal Usage: 2006 Year in Review

Friday, January 26th, 2007

It was a good year for the portal. In general, people used it exactly as we intended. They logged in, they got to WebCT, e-mail, and PeopleSoft without having to login again. In general, the portal work as a portal. Hooray, Portal!

portal-usage-2006-sm.png

Can you spot the summer months? I knew you could! So, some obvious observations:

  • Fall semester starts off with a bang. People tend to login in more than once a day during that period. This makes sense because there is a lot of schedule checking, adding, dropping, and grabbing of WebCT information.
  • Summer is slow. People rarely login more than once a day.
  • Spring break is a lot like summer, only shorter and probably not as warm.

So, what is the difference between logins and unique users? That’s an excellent question. Logins count all the times people login to the portal. If you login 3 times in one day, we count them all. To get a unique user count, we ignore how many times a person would login in a day and just count them once. This accounts for the discrepancy in our line plots. (*Note:* For the uportal geeks out there, we’re not counting guest sessions.)