Use of the PALNI Online Catalog hit record levels in October, 2006. Across all PALNI library OPACs, 9,882 unique visitors connected to the PALNI OPAC a total of 30,309 times during the month of October. Overall, these users retrieved 762,318 web pages and generated 1,325,619 hits against the OPAC web server during the month. Over the past 6 months, users have visited PALNI's OPAC web sites almost 120,000 times and displayed more than 2.8 million web pages

For the month of October, 2006, the PALNI OPAC was searched 154,116 times by keyword. The following chart shows the number of keyword searches for each library for the month of October.

A week has now gone by since our recent Aleph configuration changes and system upgrades, and PALNI OPAC performance continues to be very good. Over the past 7 days, between September 21 and September 27, keyword search response time has averaged 1.07 seconds for 40,431 keyword search across all PALNI libraries. That is a big improvement from an average of 3.54 seconds for the week of September 9 through September 15.

Individually, all libraries experienced immediate and significant improvements in response time when the system changes were implemented. For example, the following chart shows detailed, library-by-library response time data for 9/21 to 9/27, immediately after the recent system changes:

To show just how much things have improved, the next chart shows library-by-library response time data from before the system changes:

The PALNI libraries that were experiencing the most serious OPAC performance problems are seeing dramatic improvements in response time. For example, average keyword search times for Huntington University improved from almost 8 seconds before the changes to just over 1 second following the system changes. Similarly, Marian College improved from over 7 seconds to under 1 second, Butler University improved from about 6.7 seconds to under 1 second, and DePauw University improved from over 4.5 seconds to just over 1 second on average. Also note that all PALNI libraries are now well under the 3 second average keyword search response time that our contract with Ex Libris calls for.

Based on these results, and given that the system changes do not seem to have caused any new problems, it seems safe to conclude that PALNI's OPAC response time issues have been resolved at last! Staff will let Ex Libris know about the changes that we have made, since that information may be useful to other Aleph system users who are experiencing similar OPAC response time problems.

Here is the message I posted to PALNI-M on 9/21 about what is looking more and more like a major breakthrough on our long-standing OPAC response time problem. If we continue to see this kind of performance, and the changes we've made don't introduce any new surprises, we may soon be able to put our response time issues behind us.

We are collecting library-by-library statistics on OPAC response time, and should have some more detailed performance results to report by early next week.

-- 
Vince Lucas
PALNI Project Manager
INCOLSA                    |  Email:  vince at palni.edu
6202 Morenci Trail         |  Phone:  1-317-298-6570 or 1-800-733-1899
Indianapolis, IN  46268    |    FAX:  1-317-328-2380

Here are the latest keyword search response times for each PALNI library's OPAC between 8/25/06 and 8/31/06. As with the previous report, this data is extracted from the PALNI Web server log files. As such, these response times do not include any network delays, any additional time caused by a slow PC or web browser, time to execute any web page JavaScript, etc. As a practical matter, the response times on most campuses will typically be 1/2 to 1 second slower than this data indicates.

KW Search Performance for 8/25/06 through 8/31/06

Once again, we are seeing significant differences in measured response times from one campus to another. Average response times over the week ranged from around 1 second on two campuses to over 6 seconds on another, and the slowest OPACs were about 6 times slower on average than the fastest OPACs for the week.

Last week, staff implemented some minor changes to the PALNI OPAC web pages to improve our ability to collect keyword search response time data on a 'library-by-library' basis. In looking at our on-going OPAC performance problems, our perception has been that there are significant differences in keyword search performance between PALNI campuses. With the data we can now collect, we hope to confirm that our general perception has been accurate accurate, and, if so, to collect some statistics that will show the performance differences to the Ex Libris technical support team.

We have generated our first OPAC performance report from this data, for the seven day time period from August 15 through August 21. The following chart summarizes keyword search performance for the PALNI OPAC over this time period for each PALNI library.

KW Search Performance for 8/18/06 through 8/24/06

Staff implemented a new enhancement to the PALNI New Titles List this week. With this version of the New Titles List, you may limit the list of new titles that are displayed to any of eight broad subject categories. The subject classification for each new title is based on the title's LC or Dewey call number.

To use this capability, the user can select the subject category from the "Subject Area" drop down list near the top of the New Titles List web page. All of the limiting, sorting, a search options from the previous version continue to be available in this version:

For those who may not have already discovered it, the PALNI Online Union Catalog can now be searched, along with the various INSPIRE databases, through a new web portal to INSPIRE. The new INSPIRE database portal was developed using software from WebFeat, Inc., and is currently being tested on the INSPIRE web site. Click on the "Try the New INSPIREsearch" link near the top of the page.

Using the new INSPIRE web interface, users can search the PALNI Online Catalog, various other library catalogs, and any of the INSPIRE databases. The new INSPIRE portal's broadcast search capability allows the user either to search an individual database, or to run their search in parallel across many databases all at the same time.

For example, using the new INSPIRE search interface, you may search the PALNI OPAC, OCLC WorldCat, and the Indiana Shared Library Catalog (ISLC) system, all at the same time and in parallel. Similarly, the user can do a keyword search against the PALNI Union Catalog, GoogleScholar, and Ebsco's Academic Search Elite, also all in parallel.

Ex Libris recently profiled the PALNI system for OPAC performance bottlenecks, and found an Aleph V.17 problem that explains some of the slowness we've been seeing. Around January 26, they installed a temporary fix for the problem, and we have seen some improvement on our test keyword search. The test search is a simple, 2-word Boolean search that we've been running regularly over the past two years to monitor performance of the PALNI OPAC.

Since this fix was installed, average response time on our test search has improved significantly from over 5 seconds (between 1/16 and 1/22) to around 2.3 seconds (between 1/30 and 2/5). However, this is still about two times slower than we were seeing on the same search before our upgrade to Aleph V. 17.

Also, we've written software to extract actual response times from our web server log files for all OPAC keyword searches. That data shows that, between February 1 and today, Feb 8, average response time across all keyword searches against the PALNI OPAC was 4.6 secs. About 60% of keyword searches took less than 3 seconds, but more than 25% took more than 6 seconds.

We continue to have an open problem on this issue with Ex Libris. Following is the lastest PALNI OPAC performance update that we sent to Ex Libris on Feb. 8.

 

Registered users for this web site can now display overall, website-level use statistics for the PALNI OPAC. To display the latest statistical report, you must first login, then select "Support Tools" from the left-column menu, and then click on the "OPAC Web Site Statistics" link.

The statistics reports are updated at 2am each night, and provide aggregate information about total access across all the PALNI library OPAC web pages. For example, they show that, across all PALNI libraries, between October 1 and October 7, more than 3,200 unique users accessed the PALNI library OPAC websites a total of 7,400 times. Over this time period, they displayed 151,600 PALNI OPAC web pages. Peak OPAC use for the week was on Monday, October 3, and the peak hours of OPAC use were 10-11am and 2-3pm. About 60% of users stayed connected to the Web OPAC for 2 minutes or longer.

A new OPAC Web statistics page has been added to this site. Registered users to the site may display the latest statistics by selecting "Support Tools" from the left column menu, and then clicking on the "OPAC Web Site Statistics" link.

The OPAC statistics presented on this page are combined statistics for all PALNI library Web OPAC pages. Statistics collection began on September 28, so the statistics shown for now only cover a few days. However, they will grow steadily over the next month as additional OPAC use is logged.

We have had several reports of slow response time for the Aleph WebOpac. We have also noted that response time for the GUI is slow.

We have had open tickets with ExLibris abouth this issue since shortly after the switch to V. 17 production in early August. At that time, Ex Libris made some changes which corrected the extreme performance problems that we were encountering during the indexing completion. However, even after the indexing was completed, we continued to see OPAC searches that were much slower than they had been in v.15. We reported this to ExLibris in the original problem ticket and in a letter to Ex Libris' head of customer support.

On Monday, we created a new, second problem log with ExLibris to let them know that the response times were continuing to deteriorate, and we have escalated the problem to high priority. As of today, 9/13, Ex Libris has reported this problem to the second level of support (i.e., technical support) and we are awaiting their response or recommendation.

Thank you for your patience

colleen

STATUS OF BUILDS: We have had several glitches in the browse index rebuilds between Friday afternoon and today, but we have finished the last indexing job (re-started after an error suspended the job). Saturday was the nervous day, with significant errors that Ex Libris staff spent 2-3 hours off and on fixing, including repair of a couple of Oracle indexes, before he okayed me to continue with the builds.

TODAY: The browse index appears to be working correctly now, but we are NOT ready for sites to re-open it to the public on web OPAC. So far, my testing and that of Ex Libris staff show vast improvement in the browse headings.

CHECK GUI STAFF OPAC: Do start checking the browse index searching via the GUI staff OPAC. Please report any glitches with BROWSE searching on helpdesk tickets. I have turned browse searching back on for the Ancilla web OPAC and for the PALNI general OPAC so if you want to see how it is looking from the web OPAC side, feel free to check that as well.

PLEASE WAIT: Please wait for our go ahead to change your web OPACs back to allowing general browse searching by patrons. Thanks for your cooperation on this issue.

The program to fix our browse search problem was started early this afternoon, 9/2. This problem is one of the top priority open issues with our Aleph V. 17 upgrade that we have been working with Ex Libris to resolve. They estimate that it could take 60 hours of total clock time to process our entire database for this fix. As we've done in the past, to speed up the processing, we have re-assigned four additional processors and four additional GBytes of memory to Hickory to speed things up as much as possible.

We expect the program to finish sometime on Monday, and staff will be monitoring the system closely over the coming weekend for any expected problems. If everything works as Ex Libris expects, we should have the browse search problem fixed by Tuesday morning, 9/6.