Unidata Support

Through a Support Task Force (STF) composed of staff members, the UPC has been investigating ways to streamline the User Support process. The overall object of the investigation is to implement a new system which:

• simplifies and speeds the support process for all UPC staff
• provides facilities where end users can more easily help themselves
• provides a rich set of functions through which support performance can be monitored
• can be extended to provide additional functionality deemed important by UPC personnel
• seemlessly integrates with the Unidata web site

Our approach has been to identify existing support systems that, at a minimum, offer features contained in our existing inquiry tracking system and provide key features or facilities to implement key features that the STF has deemed as mandatory in a new system including the ability for the end user to open, modify and close their own inquiry, and comment on support that they have received.

After reviewing approximately 8 helpdesk packages, we decided to bring two in-house for testing:

Request Tracker
eSupport

Of these two packages, Request Tracker (RT) would be the easiest for us to extend. eSupport, on the other hand, already contains virtually all of the features that we now consider to be mandatory in a support system.

Since one of our goals is to move to a package that most helps our community, we will be soliciting input on features, ease of use, and performance for a new version of eSupport over the coming summer. Our intention is to stress test the package live both internally and externally in the soon-to-be-released upgrade to our web site, internally know as Plaza Version 2 (PV2). We will inform the community of the availability of the new support facilities through our usual channels: community email list announcements and the monthly electronic Newsletter.

Unidata Support Metrics

Above are histograms that portray the number of Unidata email responses for categories of support for a one year period ending March 29, 2005. The histograms are arranged by yearly activity averages with the highest on the left and lowest on the right. Each quarter year within the period is depicted from oldest to newest from left to right. The number of responses has been normalized to weekly averages so that the support load over the various periods can be easily compared.

Total support averaged 145 responses/week over the entire year; 143 for the first quarter; 147 for the second quarter; 143 for the third quarter; and 146 for the current quarter.

Individual support activities included in the categories shown above are listed in the following table.

Category	Packages, Groups, and Lists
ldm	ldm, ldm-users-list
idd	gozer, idd, idd-antarctic idd-brazil, idd-caribe, idd-inject, idd-status, suominet, venezuela
data	casestudies, casestudies-list, conduit, c2-linda, craft, level2-linda, craft-nws, craft-ty, datastream, difax, level2, level2-ty, noaaport, noaaport-ty
netcdf	data-models, netcdf, netcdfgroup-list, netcdf-hdf-list, netcdf-java, netcdf-perl
gempak	gempak, gembud-list
mcidas	mcdevelop, mcidas, mcidas-list
dods	dods, dods-core, dods-list, dods-tech
nws-changes	nws-changes
outreach	agu-linda, ams-linda, announce, barbados-ty, cbmet-ty, chile-ty, egrants, egy-linda, external, iai-ty, meteoforum-ty, unidata, workshop
idv	idv, idvlist, idvsteering, metapps, visad-list
utilities	decoders, ldm-mcidas, udunits
miscellaneous	misc, license, network, notrack, platforms, wxp, wxp-list
thredds	java-dev, thredds
lead	lead

Comments

• As expected LEAD support continued to grow markedly. Even so, the LEAD numbers are a major underestimate of its activity since the great majority of LEAD email transactions do not go through User Support.
• Support numbers for individual packages, groups, and lists can be seen in full support statistics.

Some tentative conclusions

• The total support provided by the UPC remains high. Overall support activities are seasonally invariant in spite the fact that support for analysis and display applications show seasonal variations.
• Support required for the legacy visualization packages GEMPAK and McIDAS continues to be substantial with GEMPAK leading the way as the most widely distributed analysis application. The IDV support load has grown substantially as more users adopt the package.
• Support required for LDM, IDD, and data continues at a high level and shows some variability throughout the year.
• Support for netCDF remains high.
• The overall higher numbers for outreach reflect inclusion of more categories of support/outreach provided by Linda.
• Taken as a whole, the support required for visualization packages (GEMPAK, IDV, and McIDAS) is comparable to the support related to data reception (LDM, IDD, noaaport).
• Support for any package increases after a new version of the package is made available.

NOTE

These numbers and conclusions should not be taken too literally, for several reasons:

• For some packages, multiple responses in the same thread may be bundled into a single archived email. Other packages have each response in a thread counted separately.
• After a new release of software, there may be a flurry of the same or similar questions, which can be answered in separate emails or in a single mailing list posting.
• This counts primarily support of end users and site administrators, not developers. Support for other developers in projects such as OPeNDAP/DODS, THREDDS, McIDAS, GEMPAK, and IDV requires significant resources, but is difficult to assess.
• Not all support records were indexable for this report. Given this, the above numbers are an underestimate of the actual support being provided by the UPC.

by Tom Yoksas