ITPAC Meeting Minutes: May 8, 2008
Last edited May 20, 2008 by Steve Lasley.
- Jump to Items for Administrative Notice.
- Download and print Meeting Agenda [26KB doc].
- Download and print "Google Analytics vs. Urchin on the Same Set of Web Sites" handout [116KB pdf].
- Listen to Streaming Audio archive of the meeting.
Agenda
- Approve Prior Minutes
- Administrative Updates
- IT Updates
- ICC Updates
- Website Analytics Policy
- Web and Domain Policies
- Other Discussion
Call to Order
This meeting was scheduled in McCarty D, room 1031 at 9:00am on Thursday, May 8, 2008. The meeting was called to order by chairman Allen Wysocki.
Attendees
Ten people attended this meeting locally with one attendee via Polycom videoconference.
Ten members were present: Dan Cromer, Steve Johnson (via Polycom), Joe Joyce, Steve Lasley, Sheri Munn, Dave Palmer, Michelle Quire, Wendy Williams, Ashley Wood and Allen Wysocki.
Five members were absent: Keith Gouin, Brian Gray, Mary Anne Morgan, Brandon Segermeister and Pete Vergot.
One visitor was present as well: Jan van der Aa.
Audio Archive
An archive of audio from the meeting is available.
1.) Approve Prior Minutes
The minutes of the last meeting [33KB doc] were approved without dissent.
2.) Administrative Updates
Joe Joyce spoke first because he had to leave for an important videoconference. He reported that IFAS hopefully is in final negotiations with Florida Crystals Corporation for the signing of a $20 million contract to build a cellulosic ethanol plant.
2.1) Budget Cuts
2.1.1) The budget cuts were almost much worse
Joe said he knew that the budget was on everyone's mind. The $9.5 million cut which IFAS took would have been closer to $25 million had the agricultural community not stepped up on our behalf.
2.1.2) Who had input?
Some concern has been expressed that more people should have been involved in discussions on how the cuts were distributed within IFAS. The VP's office had talked with the Faculty Council, unit heads, the deans, and some of our clientele. That really is about as far as they could go with it.
2.1.3) How were cuts distributed?
Each unit was allocated a reduction based on their total unfilled and filled positions. Much was taken off the top as well. Seventeen unfilled positions were eliminated saving about $7 million. There are another fourteen people that are leaving between now and 2010 for a variety of reasons; those positions will not be refilled. Money was also taken out of central operations.
2.2) Reorganization
2.2.1) A move toward web-based communications
IT, ICS and EMR are being merged. This is not happening due to past performance, but rather because IFAS administration believes we need to look at moving more toward Web-based communications. Included in that is Distance Education, whether it be extension certification or some form of short-course, as well as formal undergraduate coursework. This group is now going to focus on how we design those sorts of communications.
Joe reported that Ashley Wood, Dan Cromer, and Jack Battenfield have met with him and Larry Arrington twice on this issue and those three have had four other meetings as well. Their goal is to develop a structure that is leaner than before (as each group has taken cuts) and can shift our communications from print to the Web. Those three will bring a report back to a committee consisting of many folks across IFAS who will review that to see if it meets our needs.
2.2.2) Programming staff reduced - WAN moving to CNS
Joe reported that programming staff had been decreased and the Wide Area Networking (WAN) is moving from being an IFAS operation to a CNS/central university operation. That has resulted in a series of position eliminations and those folks have been notified. By July 1st, Dan Cromer will have a letter of agreement from CNS on how the WAN relocation will operate. That is expected to save us about $35 thousand a year.
2.2.3) ICS is weathering the storm
Ashley Wood added that ICS was fortunate to be able to shift personnel onto revenue accounts as a short-term solution to their portion of the budget cuts. He believes that continuing high sales levels might permit turning that into the long-term solution as well. Ashley reported that sales of identity products have increased in conjunction with the recent support demonstrated on behalf of IFAS by our community.
2.3) Tremendous Support from IFAS Clientele
The broad spectrum of support for IFAS has been remarkable.
Joe added that support for IFAS in these troubled times has been incredible. This came not just from commodity agriculture but from our entire clientele base including 4-H, FYCS and the environmental community that rely on us for research and extension information. These folks came together for us as a group more cohesively than anything we have observed in the last twenty to thirty years. Instead of each group trying to cut their own best deal, they all hung together and demonstrated incredible effort for which IFAS is exceedingly grateful.
2.4) Could budget cut allocations still be altered?
Allen Wysocki mentioned that he is on the UF Faculty Senate Steering Committee and has seen an enormous volume of e-mail regarding the cuts. There are at least one hundred pages of e-mails from CLAS faculty to the President. Allen asked Joe if there was any chance that budget cut allocations might change as a result of the uproar?
Joe admitted that he didn't know. The point to remember for IFAS, however, is that we started out facing the possibility of an inordinately large cut and ended up with the same 6% reduction as the rest of the university.
2.5) Concerns Over the WAN Move
Ashley mentioned concern about moving the WAN operation simply because IFAS is different than the rest of the university. He gave the past example of the IFAS mail room. He believes moving that to the central facility worked so well partly because we relocated two IFAS staff centrally as part of that exchange. Ashley asked if that is what may happen with the Wide Area Network, as he feels we need somebody at CNS who has experience working with IFAS.
Dan Cromer responded that it is the intent of CNS to advertise and hire a new position. CNS has already handled fifteen of our circuits, so it is not as if they haven't been involved all along. Each of the REC sites has been handled by CNS for the last couple years. Dan does have some concerns that wiring projects which central IFAS IT previously would have handled internally will now need to be contracted out. On the other hand, Dan has been trying to get out of that business since 2002 when two wire-installer positions were lost due to cuts. Dan, however, does have great confidence in CNS for handling our networks.
Joe added that Dan already knows that he will have to remain as watchdog over CNS to ensure that we get what we are paying for. Joe stated that IFAS IT has traditionally been ahead of the rest of the university. In those areas where UF has caught up, such as with Exchange, we have generally made the decision to move functions centrally.
3.) IT Updates
3.1) Consolidation
Dan began by saying that Dr. Joyce had already covered the main two topics he had planned to mention. The first is that Dan is working on a consolidation plan with Ashley and Jack Battenfield and they have refined that to a point where at least Joe Joyce and Larry Arrington seem to be happy with it. They will eventually take that report to the larger committee for further consideration.
3.2) IT Staff Layoffs
Regarding the layoffs, Dan reported that central IFAS IT had to lay off five staff members: Petraq Papajorgji of the software group, Bill Cope of the business software group, Bill Lancaster of the Help Desk, as well as Jennifer Xu and Claude King both from the WAN group that is being outsourced to CNS.
3.2.1) Two retiring staff positions will not be refilled
Additionally, Dan reported that Tom Hintz will retire at the end of November and Ann Hutcheson by January. Their positions will not be refilled.
3.2.2) Service levels bound to suffer
Consequently, IT will be struggling to provide the same level of overall service and Dan does not believe they really can. Dan believes the loss of the programmer positions will be particularly hard felt.
3.2.3) Unit-level staff attrition will increase demands
Steve Lasley added that at least one unit IT staff person has been lost as well, and that will place additional support requirements on the central group.
3.2.4) Other upcoming staff changes
Other non-layoff staff changes include Daniel Halsey (programmer who is 25% IFAS and 75% SARE) and Mark Ross (IT/SA web server administrator). Both will be leaving shortly. Daniel's position has been approved for refilling and Dan believes Mark's position will certainly have to be refilled as well.
3.2.5) Livestock Pavilion to be vacated
At Joe Joyce's request they also plan to vacate the Livestock Pavilion (Bldg 66), relocating offices currently housed there to either Bldg 120, Bldg 87 or Bldg 162.
4.) ICC Updates
4.1) A Recommendation for Barracuda Management
Steve Lasley passed out copies of an ICC recommendation handout [36KB doc] noting that members had been pointed to a more useful web-based version (due to its embedded links) via prior e-mail. Steve began by admitting that overall the anti-spam measures taken at UF are enormously complex. In this document, Steve has attempted to simplify things a little bit in order to address one particular aspect of importance.
4.1.1) Background
Prior to moving our Exchange service to UF, e-mail that reached the Exchange server from outside was not blocked; rather, it was delivered either to the Inbox or the Junk E-mail folder. Now, with the Barracuda anti-spam appliance, there is a considerable amount of blocking taking place at this level--some of which is under user control.
4.1.2) Quarantine more and block less
Our concern, and the reason for this recommendation, is that the ICC feels it would be useful to have the default controls set as closely as possible to what we had been used to prior. This recommendation suggests that we lighten-up on the blocking and place those instead into quarantine where the user will be able to see them and possibly catch something which was inappropriately scored as spam.
4.1.3) Affect on quarantine levels
The main concern with moving from blocking to quarantine would be that those lists would grow so long that people would not have/take the time to check for false positives. Our investigations, however, suggest that this increase would be very minimal on average.
4.1.4) Remediation
We realize that remediation will have to be available for atypical users and that IT staff would have to be right there to assist in changing things back individually for those for whom the new defaults proved inappropriate.
4.1.5) The next step towards implementation
The ICC has handed this recommendation off to Mark Rieger, the IFAS representative on the UF Exchange Advisory Committee, who will attempt to get it implemented at the UF level. Mark has already raised this issue to the committee at their last meeting and the ICC rewrote its recommendations somewhat in response to Mike Conlon's comments at last month's ICC meeting.
The reason Dan Cromer wanted the ICC to bring this matter to ITPAC was two-fold. Firstly, agreement from this committee would add support to Mark Rieger's negotiations. Secondly, whether or not UF Exchange is persuaded by our argument, we would still want these to be documented as the recommended settings for IFAS.
4.1.6) Questions & Answers
Q: Allen Wysocki: Has this been precipitated by concerns of people having had e-mails blocked?
A: There have been documented instances of e-mail being blocked which should have been delivered. It is difficult, however, for an individual to document what that don't receive. Initially, we were able to acquire some logs which indicated what was being blocked overall; that led to discussions concerning which messages might or not might constitute spam--the end result being that this required the intended recipient to make that interpretation.
We have not raised the matter to our general public, however, because we do not want people going in and changing their settings "willy-nilly". It would be very easy for people to go into the settings and make changes there which would make things worse for everybody. Although a web site has been created documenting how to change settings to match these recommendations, our hope is rather that instead we can be successful in having the defaults set for our users per this recommendation:
Q: Allen Wysocki: Why do I keep having to whitelist e-mails from a particular organization?
A: This is something we need to get documented as well. When you whitelist a particular e-mail message the entire address gets entered into that whitelist entry. Since many newsletters and the like come from basically random rotating addresses within a particular domain, what one needs to do is go into the Preference > Whitelist configuration under their account within the Barracuda web interface and enter just the portion of the address which follows the "@" symbol--i.e., the domain from which the message originated. There is an off-setting complication, however, in that some domains may include both addresses that you do and do not wish to receive messages from.
The Barracuda is very good at what it does and, for most people, once you get your important contacts properly whitelisted you probably won't get anything in your quarantine which was caught inappropriately.
Q: Michelle Quire: So following these recommendations would increase the number of messages we would see in our quarantines?
A: Yes, but on average the numbers would be very small; the average number of e-mails quarantined daily per user would increase from only 0.46 to 0.53. It just depends on the user. The distribution of score-blocked messages across users is much skewed with the majority of users getting very few daily. You should realize that we are only talking about e-mail entering our system from the outside. E-mail internal to the UF Exchange organization does not pass through the Barracuda.
Q: Dave Palmer: Will changing the defaults affect users who have already changed their settings?
A: No. Even if the defaults are changed, users who have already customized their settings will not be affected. Though it is somewhat premature to even mention, there may be a way to go back and script particular settings for a group of individuals. That method might, for example, be used to set things back for those whom it was determined would suffer from greatly increased quarantine sizes under our recommendations.
4.1.7) Further Comments
Michelle Quire mentioned missing many messages from international students who tried to contact them. Steve responded that there is a lot of blocking going on at the UF level as well. That blocking, however, is RFC compliant in that the sender receives a bounce message indicating that the message was not delivered. When something is blocked by the Barracuda, although it is retained there, no one knows; the sender doesn't get a bounce and the intended recipient gets no indication that something was blocked. The only way you can retrieve such a message is to contact the Exchange admins and have them pull the logs and look at them--something which is very time consuming.
There is also the fact that 86% of the blocking which the Barracuda does is not based on message scoring; thus there is a lot of blocking which the user has no control over in any case. That sort of blocking is done through various measures which intend to make quite certain that these are things which you would not wish to receive. It is not that the Barracuda is not a good device, but rather that it could be tuned better for our purposes. In general though, we do not want to reset things for all because we need to assume that any customizations individuals might have made were both intentional and desired; that might not be the case, but we would have to assume that.
Dave Palmer mentioned that the Barracuda merely addresses spam after it has already arrived. He believes there are things we could be doing on the front-end to help reduce spam. One example is to not publish readily harvestable e-mail addresses on our various web sites. Steve responded that, while this is very true, such a discussion would be outside the scope of this particular recommendation.
4.1.8) Motion to approve the recommendation of the ICC
Dan Cromer made the motion, Ashley Wood seconded it, and the motion passed unanimously.
5.) Website Analytics Policy
5.1) Background
Dave Palmer provided a handout [116KB pdf] along with some background information. He related that several years ago IFAS administration asked for and negotiated a funding formula with the Florida Legislature called the "RDU" formula. In essence it says that if IFAS Research and Extension do more business, then IFAS gets more funding. Unfortunately, the formula that was negotiated at the time does not include electronic contacts at all.
Note: The IFAS funding formula includes five measures:
- Educational materials prepared (an original work such as a journal article, book chapter, fact sheet, or handout, but excluding newspaper articles and TV productions)
- Field consultations
- Office consultations
- Telephone consultations
- Group learning participants (clients present during workshops, meetings, seminars, field days, interactive video conferencing etc.)
Dave's understanding from discussions with Larry Arrington is that IFAS plans to go back to the legislature and ask that the formula be altered to include several different types of on-line contacts.
5.2) IFAS Requires Defensible Analysis Methods
The problem is that tracking on-line contacts will require a relatively accurate tool and the process will require procedures and likely policies as well. At this time, even though we do have a single supported tool, there is no policy that says everyone has to use it.
Across IFAS, Dave knows of at least two different tools that are being used to track web sites. Urchin is the log-based/server-based tool that is supported currently by IFAS, and Dave has utilized a page-based/script-based tool on his sites called Google Analytics. The handout he provided shows a roughly three-fold difference in the page visits reported using Google Analytics vs. Urchin on the same set of web sites. Dave made the case that the disparity demonstrated between these two methods raises a question regarding the relative credibility of the tools we currently use.
Allen Wysocki asked if methods used by other states had been investigated. Dave responded that Marion Douglas had performed a survey of extension IT at various locations across the country and found that every conceivable tool was being used and that there was no single tool which was considered to be the standard for comparison. The disparity between the two tools which Dave has investigated is difficult to account for, however, because both use proprietary algorithms.
5.3) When and How Should IFAS Proceed?
Dave stated that IFAS would eventually need a policy on this and we are going to have to settle on a tool. The question he raised to ITPAC was how do we wish to handle that? Do we want to be proactive and begin to address this now or do we want to wait and be reactive later? How do we deal with this so we can give administration a relatively defensible tool? Once a tool was selected we would still need to convert IFAS to using that and begin collecting data which administration may use in their efforts to get electronic contacts included into our funding formula.
Ashley Wood agreed that we need an accurate and defendable solution and said that Diane Craig, Coordinator of Statistical Research for PDEC, was coming to his office the very next day to discuss this same thing. Her office is being asked about these issues and she is the one who will need to provide Larry Arrington with reports that he can use to prove that our resources are valuable to our clientele. She has been asked to develop a tool to replace Unifas for this process. How do we prove that we have one million visits a month on EDIS? Ashley has heard those numbers repeatedly, but doesn't know that to be true. How do we know that the unique visits for Solutions for Your Life continue to go up? We have a method of looking at that and saying that they do--but is it provable?
5.4) Clarification on the What's and How's of Tracking
Sheri Munn asked for clarification about how a web site owner goes about utilizing a particular tracking method. Dave explained that Urchin is on the server and can be used to analyze a site without changing the web site itself. Using Google Analytics involves adding code to each page on a web site. Sheri then asked if our various IFAS web sites were then using one or the other. Dave pointed out that, for the sites listed in his handout, both methods were in effect; that is how he was able to obtain that comparison.
Allen Wysocki asked if the issue was what to track or rather how to track it. Dave responded that both of those were valid issues.
5.5) Two Basic Tracking Methods
Dan Cromer provided an explanation that compared and contrasted the two basic methods used for tracking.
5.5.1) Log-based/server-based methods
On any web server there is a process running that listens to the network traffic and logs every single connection. These logs are very verbose and note access at a very granular level. If you have a web site that has ten pictures, for example, a separate log entry will be made for the access to each of those items as the page is loaded. These entries are made for each item on each page that is visited. The logs exist on the server that is providing the web pages, however, so there is no possibility of a network problem messing up the statistics. Currently, with our use of the Urchin tool, we mine the data which is kept on our own servers within their logs.
5.5.2) Page-based/script-based methods
Page-based tools require that you place code directly within each web page. That code communicates with a remote service over the network where the statistics are maintained. With Google Analytics, it is Google that is keeping and maintaining those statistics--not us.
5.5.3) All methods vary in their results
There are numerous products (like Urchin) which mine web server logs to turn that raw data into useful information. Urchin is the tool we use currently, but we had used DeepMetrix prior. Even those two solutions, both server-based, provided different results with the same data.
It is clear that all the available tools are better at detecting trends over time within a particular product than in estimating exact numbers of visits. Dave pointed out that, for various detailed technical reasons, accurately tracking "visits" is not an exact science. As each tool uses a different technique, each will produce a different result.
5.6) Server-based Methods Deemed Superior
Dan said that he much prefers the log-mining method--whichever product is used. The data is already being maintained via the server logs; that allows you to go back and do historical analysis. A script-based method would require that code be added and no statistics would be available prior to doing that. Steve added that a page-based approach would also require a content management system to ensure that each page of each web site was correctly coded.
Wendy Williams said that she felt there was a far better chance of getting good data from a server-based product as well since it is using the best information source available--the server logs. Ashley Wood said that his discussions with Marion Douglas and Diana Hagan indicated that they concur.
5.7) We Must First Focus on What Data to Collect
Steve Johnson said he believes we should first ask what data it is that we want to collect. Secondary to that is the determination of what product can do that at a price which is affordable.
Ashley agreed that we really need to go back and look at what it is we want to collect and why. He mentioned that when they start putting together the extension calendar, the process they use will involve looking at Solutions for Your Life and EDIS and noting what information it is that people are seeking. That will tell them how to create the calendar so that each topic area will bring people back to the web site that is being demanded. That is just one example of the sorts of information IFAS needs to collect. There is much that needs to go back to Department Chairs and the Deans as well so they can demonstrate to the legislature that the information IFAS provides is valuable and being used.
5.8) We should also look at what others have done
Ashley Wood suggested that we also might look into how other organizations, such as eXtension, were collecting statistics. Michelle Quire asked if any other states have already reached such an agreement with universities. If so, their experiences would be very valuable in learning what methods are deemed acceptable.
5.9) ITPAC Requests the Creation of Sub-committee on Web Analysis
Ashley Wood proposed the following motion:
ITPAC recommends that a sub-committee be created to investigate IFAS web analysis needs. It should first determine what data we need to track and secondly how we should implement tracking. The solution needs to be defendable to the legislature and useful in promoting the value of the services IFAS provides. This sub-committee should prepare a report for the consideration of ITPAC and should be comprised of those most concerned with this data and how it is collected. We recommend that the committee include Dave Palmer, Wendy Williams, Diana Hagan, Marion Douglas, Mark Ross, Diane Craig and Brian Gray.
The motion was seconded and passed unanimously.
6.) Web and Domain Policies
Ashley Wood brought up a subject mentioned briefly at the last ITPAC meeting and discussed in detail at last November's ITPAC meeting, namely the status of the UF/IFAS Web Policy [68KB doc] and UF/IFAS Domain Name Policy [72KB doc] proposals.
6.1) Web Analysis is Linked to the Domain Policy Issue
Ashley suggested and Dave Palmer agreed that these policy proposals had some tie-in to the web analysis issue just discussed. Dave is concerned that Urchin may not be able to track statistics on folders--something that IFAS will require if our web sites are to be consolidated as per the proposed Domain Policy.
6.2) Procedures related to these policies are already in place and being used
Ashley stated that the standards introduced by the proposed Web and Domain Policies have already made their way into practice. He also knows that the consolidation espoused by the Domain Policy is something which Mark Ross is very interested in seeing happen.
6.3) Overview of what issues the policies cover
Allen Wysocki asked for a synopsis of what these policy proposals covered. Ashley provided some of the history of their development and Dan Cromer spoke briefly on the rationale behind the web site consolidation plan. Ashley pointed out that the Web and Domain Policies are separate though related documents. The Domain Policy relates more to the consolidation which Dan spoke about.
The Web Policy states that IFAS needs to have a uniform global approach to the Web. The Web Policy states that before you may have a web site you must meet certain criteria regarding its purpose and upkeep. It also provides for content and technical reviews to ensure sites are properly maintained. The intention is not to be content police, but rather to provide a resource that can be used to help maintain and improve the overall IFAS web presence.
6.4) Website consolidation concerns
Dave Palmer reiterated his concern regarding web site consolidation--namely that he isn't sure we can acquire the necessary statistics for freestanding sites once they are moved into folders beneath larger sites.
Steve Lasley responded that we do need to be able to track what is in folders, but we also very much need to consolidate. To make our web presence manageable we have to eventually have some kind of content management system (CMS), and what we have currently cannot be poured into a CMS. That is why the domain restructuring is so important. If we can start moving in the right direction then we are going to be so much further ahead when we get to the point where we have a CMS product and are looking at "this" and putting it "there". That is why Steve has backed the consolidation so strongly even though it is painful and people are not going to like all aspects of it. That is still the direction in which we simply must move.
6.5) ITPAC recommendation for publishing Web Policy as IMM
Ashley said that he would like a recommendation from ITPAC to accept the proposed Web Policy and implement it as an IMM. Steve Lasley made that motion and Ashley seconded it. During follow-up discussion Michelle Quire asked if we were certain that there was no mention of the domain consolidation issue within the Web Policy document. Sheri Munn responded that she hadn't read either of these policy documents. Dan Cromer then suggested that the Web Policy document be revised by Steve Lasley and/or Ashley Wood and be rewritten as an IMM [48KB doc]. If there are any references to consolidation those will be removed in the process. The resultant proposed IMM would then be distributed back to the committee who would vote on the matter via e-mail. Assuming it passed, it would then be forwarded to the VP's office for their consideration.
6.6) Domain Policy to be revisited by Web Analysis sub-committee
It was also decided that the Domain Policy would be passed along to the proposed Web Analysis Sub-committee for continued consideration as part of their charge. Dave Palmer would like the proposed Web Analysis sub-committee to be ready to come back with some form of report by the next ITPAC in August--perhaps even being ready with a policy recommendation.
7.) Other Discussion
7.1) Other new IT projects at UF
Allen Wysocki asked Dan Cromer if there were other IT changes happening centrally.
Dan reported that more committees are being formed. Mike Conlon is working on giving an "IT grade" to UF.
Jan van der Aa added that this effort comes from the annual list which Educause is producing. We want to score how UF is doing in keeping up with various IT-related tasks such as security and identity management.
Another initiative is looking at generating policies concerning building classification and who provides the network infrastructure in various areas across UF.
Next Meeting
The next regular meeting is tentatively scheduled for August 2008.
Items for Administrative Notice
- ITPAC unanimously approved the ICC Recommendation for Barracuda Management.
- ITPAC Requests the Creation of Sub-committee on Web Analysis.
- The proposed Web Policy document is being rewritten as an IMM [48KB doc] for consideration by the VP's office.