Re: [U2] Large DICT affecting I/O
So, how many reports or queries do you have that reference all 2000 dict items ... really? When you are working with this 2,000 field item, do you work with the item as a dimensioned array, or a dynamic array? How many of these 2,000 data elements are multi-valued (sub-multi-valued? or worse??) and how many of these are associated multi-values Performance issues are NOT related to the dictionary size (though dictionaries may accumulate debris over time), as the dictionary can be readily resized to be the right size -- I would hypothesise that if you had the tools to perform a field level data analysis of USAGE (how many fields appear grouped on screens reports together, how often these are used) you would find that there are some hot fields that are heavily used every day, whilst others are used sporadically every week/month, which could be used to guide any re-engineering. However, this is probably all academic, because in reality you are probably not in a position to re-structure the database anyway :-( Ross Ferris Stamina Software Visage Better by Design! -Original Message- From: u2-users-boun...@listserver.u2ug.org [mailto:u2-users-boun...@listserver.u2ug.org] On Behalf Of Jeffrey Butera Sent: Wednesday, August 7, 2013 8:07 PM To: U2 Users List Cc: U2 Users List Subject: Re: [U2] Large DICT affecting I/O Thanks to those who replied. 1) we have tools to easily edit dictionaries of any size (one of the few benefits of Datatel/ Ellucian) 2) we have tools to monitor and resize dictionaries just like any other file ( also a Datatel benefit) It's more of a general question about performance. We import data from the Common Application which has almost 2000 data attributes per person. Thus having a large DICT isn't sloppy or lazy work on our end, it's a necessity of the data. Thus the question is better stated as: Is unidata performance better if we stuff all 2000 elements in a single DICT or break the data into multiple (eg: 4) files of 500 elements each? When we work with this data we need all 2000 elements so is reading 4 or 5 separate tables any more efficient than reading a single large table of 2000 elements? Jeffrey Butera, PhD Associate Director for Application and Web Services Information Technology Hampshire College 413-559-5556 On Aug 6, 2013, at 11:16 PM, Doug Averch dave...@u2logic.com wrote: Hi Jeffery: We have a client with 6,000 dictionaries items and they have no performance problems If the dictionary is sized correctly, there generally is no performance hit. However, editing it with some tools is a pain because it takes quite a long time to read them. Regards, Doug www.u2logic.com XLr8Dictionary Editor for large dictionary editing On Tue, Aug 6, 2013 at 9:08 PM, jeffrey Butera jbut...@hampshire.eduwrote: I'm curious how large of a DICTionary some of you have worked with and, in particular, how very large DICTs can adversely affect applications. We have a DICT approaching 1500 data elements (no idescs) - which is quite large for us. But I'm curious if others have DICTs this large or larger and have no adverse affect on their application performance. This is Unidata 7.3.4 if it matters. -- Jeffrey Butera, PhD Associate Director for Application and Web Services Information Technology Hampshire College 413-559-5556 __**_ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/**mailman/listinfo/u2-usershttp://listser ver.u2ug.org/mailman/listinfo/u2-users ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users
Re: [U2] Large DICT affecting I/O
Jeff By 'large dictionary' do you mean that the dictionary itself is too large - has lots of synonyms - or that the data it is describing has that number of fields and so has become too large for efficient storage? If the former, I've found people often forget to resize their dictionaries and the VOC file alongside the data. VOC is particularly vulnerable as everything goes through it. I've seen a VOC file with half a million entries in it on one site. Dictionary and VOC are no different storage wise to other files, they need to be cared for :) In terms of the records being described, however, that's more of an issue. Are you getting efficient storage? If data is being prematurely pushed into overflow - even level 1 - that's bound to cause performance issues. And UniData doesn't have the hint mechanisms of UniVerse so I'd suspect that accessing higher order fields would be slow, though I've not benchmarked that. Brian -Original Message- From: u2-users-boun...@listserver.u2ug.org [mailto:u2-users-boun...@listserver.u2ug.org] On Behalf Of jeffrey Butera Sent: 07 August 2013 04:09 To: U2 Users List Subject: [U2] Large DICT affecting I/O I'm curious how large of a DICTionary some of you have worked with and, in particular, how very large DICTs can adversely affect applications. We have a DICT approaching 1500 data elements (no idescs) - which is quite large for us. But I'm curious if others have DICTs this large or larger and have no adverse affect on their application performance. This is Unidata 7.3.4 if it matters. -- Jeffrey Butera, PhD Associate Director for Application and Web Services Information Technology Hampshire College 413-559-5556 ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users
Re: [U2] Large DICT affecting I/O
Thanks to those who replied. 1) we have tools to easily edit dictionaries of any size (one of the few benefits of Datatel/ Ellucian) 2) we have tools to monitor and resize dictionaries just like any other file ( also a Datatel benefit) It's more of a general question about performance. We import data from the Common Application which has almost 2000 data attributes per person. Thus having a large DICT isn't sloppy or lazy work on our end, it's a necessity of the data. Thus the question is better stated as: Is unidata performance better if we stuff all 2000 elements in a single DICT or break the data into multiple (eg: 4) files of 500 elements each? When we work with this data we need all 2000 elements so is reading 4 or 5 separate tables any more efficient than reading a single large table of 2000 elements? Jeffrey Butera, PhD Associate Director for Application and Web Services Information Technology Hampshire College 413-559-5556 On Aug 6, 2013, at 11:16 PM, Doug Averch dave...@u2logic.com wrote: Hi Jeffery: We have a client with 6,000 dictionaries items and they have no performance problems If the dictionary is sized correctly, there generally is no performance hit. However, editing it with some tools is a pain because it takes quite a long time to read them. Regards, Doug www.u2logic.com XLr8Dictionary Editor for large dictionary editing On Tue, Aug 6, 2013 at 9:08 PM, jeffrey Butera jbut...@hampshire.eduwrote: I'm curious how large of a DICTionary some of you have worked with and, in particular, how very large DICTs can adversely affect applications. We have a DICT approaching 1500 data elements (no idescs) - which is quite large for us. But I'm curious if others have DICTs this large or larger and have no adverse affect on their application performance. This is Unidata 7.3.4 if it matters. -- Jeffrey Butera, PhD Associate Director for Application and Web Services Information Technology Hampshire College 413-559-5556 __**_ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/**mailman/listinfo/u2-usershttp://listserver.u2ug.org/mailman/listinfo/u2-users ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users
Re: [U2] Large DICT affecting I/O
Jerry, As previous posts have pointed out a Dictionary is no different than any other file. It need to be properly sized for the data it holds. When properly sized will have maximum performance by definition. No need to 'split' files At TCL: 'file.stat DICT filename' Example: 'file.stat DICT PARTS' If file.stat has a new file size recommendation on its last line then run (after hours with no users or running processes that may use the dictionary): At TCL: '!memresize DICT filename new.size' Example: '!memresize DICT PARTS 1009' Marc Rutherford Principal Programmer Analyst Advanced Bionics LLC 661) 362 1754 -Original Message- From: u2-users-boun...@listserver.u2ug.org [mailto:u2-users-boun...@listserver.u2ug.org] On Behalf Of Jeffrey Butera Sent: Wednesday, August 07, 2013 3:07 AM To: U2 Users List Cc: U2 Users List Subject: Re: [U2] Large DICT affecting I/O Thanks to those who replied. 1) we have tools to easily edit dictionaries of any size (one of the few benefits of Datatel/ Ellucian) 2) we have tools to monitor and resize dictionaries just like any other file ( also a Datatel benefit) It's more of a general question about performance. We import data from the Common Application which has almost 2000 data attributes per person. Thus having a large DICT isn't sloppy or lazy work on our end, it's a necessity of the data. Thus the question is better stated as: Is unidata performance better if we stuff all 2000 elements in a single DICT or break the data into multiple (eg: 4) files of 500 elements each? When we work with this data we need all 2000 elements so is reading 4 or 5 separate tables any more efficient than reading a single large table of 2000 elements? Jeffrey Butera, PhD Associate Director for Application and Web Services Information Technology Hampshire College 413-559-5556 On Aug 6, 2013, at 11:16 PM, Doug Averch dave...@u2logic.com wrote: Hi Jeffery: We have a client with 6,000 dictionaries items and they have no performance problems If the dictionary is sized correctly, there generally is no performance hit. However, editing it with some tools is a pain because it takes quite a long time to read them. Regards, Doug www.u2logic.com XLr8Dictionary Editor for large dictionary editing On Tue, Aug 6, 2013 at 9:08 PM, jeffrey Butera jbut...@hampshire.eduwrote: I'm curious how large of a DICTionary some of you have worked with and, in particular, how very large DICTs can adversely affect applications. We have a DICT approaching 1500 data elements (no idescs) - which is quite large for us. But I'm curious if others have DICTs this large or larger and have no adverse affect on their application performance. This is Unidata 7.3.4 if it matters. -- Jeffrey Butera, PhD Associate Director for Application and Web Services Information Technology Hampshire College 413-559-5556 __**_ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/**mailman/listinfo/u2-usershttp://listser ver.u2ug.org/mailman/listinfo/u2-users ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users
Re: [U2] Large DICT affecting I/O
Provided your dictionaries, in the one file scenario, or the five file scenario, are each and all, properly sized THEN You are more efficient in reading all dict entries from a single dict file BECAUSE Your app is only handling a single file pointer AND Your internal refs are only viewing groups from a single file and registers are only pointing at groups in a single file So your internal usage pool is much smaller, which also means less internal clean up work when you are done -Original Message- From: Jeffrey Butera jbut...@hampshire.edu To: U2 Users List u2-users@listserver.u2ug.org Cc: U2 Users List u2-users@listserver.u2ug.org Sent: Wed, Aug 7, 2013 3:07 am Subject: Re: [U2] Large DICT affecting I/O Thanks to those who replied. 1) we have tools to easily edit dictionaries of any size (one of the few benefits of Datatel/ Ellucian) 2) we have tools to monitor and resize dictionaries just like any other file ( also a Datatel benefit) It's more of a general question about performance. We import data from the Common Application which has almost 2000 data attributes per person. Thus having a large DICT isn't sloppy or lazy work on our end, it's a necessity of the data. Thus the question is better stated as: Is unidata performance better if we stuff all 2000 elements in a single DICT or break the data into multiple (eg: 4) files of 500 elements each? When we work with this data we need all 2000 elements so is reading 4 or 5 separate tables any more efficient than reading a single large table of 2000 elements? Jeffrey Butera, PhD Associate Director for Application and Web Services Information Technology Hampshire College 413-559-5556 On Aug 6, 2013, at 11:16 PM, Doug Averch dave...@u2logic.com wrote: Hi Jeffery: We have a client with 6,000 dictionaries items and they have no performance problems If the dictionary is sized correctly, there generally is no performance hit. However, editing it with some tools is a pain because it takes quite a long time to read them. Regards, Doug www.u2logic.com XLr8Dictionary Editor for large dictionary editing On Tue, Aug 6, 2013 at 9:08 PM, jeffrey Butera jbut...@hampshire.eduwrote: I'm curious how large of a DICTionary some of you have worked with and, in particular, how very large DICTs can adversely affect applications. We have a DICT approaching 1500 data elements (no idescs) - which is quite large for us. But I'm curious if others have DICTs this large or larger and have no adverse affect on their application performance. This is Unidata 7.3.4 if it matters. -- Jeffrey Butera, PhD Associate Director for Application and Web Services Information Technology Hampshire College 413-559-5556 __**_ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/**mailman/listinfo/u2-usershttp://listserver.u2ug.org/mailman/listinfo/u2-users ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users
[U2] Large DICT affecting I/O
I'm curious how large of a DICTionary some of you have worked with and, in particular, how very large DICTs can adversely affect applications. We have a DICT approaching 1500 data elements (no idescs) - which is quite large for us. But I'm curious if others have DICTs this large or larger and have no adverse affect on their application performance. This is Unidata 7.3.4 if it matters. -- Jeffrey Butera, PhD Associate Director for Application and Web Services Information Technology Hampshire College 413-559-5556 ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users
Re: [U2] Large DICT affecting I/O
Hi Jeffery: We have a client with 6,000 dictionaries items and they have no performance problems If the dictionary is sized correctly, there generally is no performance hit. However, editing it with some tools is a pain because it takes quite a long time to read them. Regards, Doug www.u2logic.com XLr8Dictionary Editor for large dictionary editing On Tue, Aug 6, 2013 at 9:08 PM, jeffrey Butera jbut...@hampshire.eduwrote: I'm curious how large of a DICTionary some of you have worked with and, in particular, how very large DICTs can adversely affect applications. We have a DICT approaching 1500 data elements (no idescs) - which is quite large for us. But I'm curious if others have DICTs this large or larger and have no adverse affect on their application performance. This is Unidata 7.3.4 if it matters. -- Jeffrey Butera, PhD Associate Director for Application and Web Services Information Technology Hampshire College 413-559-5556 __**_ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/**mailman/listinfo/u2-usershttp://listserver.u2ug.org/mailman/listinfo/u2-users ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users
Re: [U2] Large DICT affecting I/O
Hi Jeffery: We have a client with 6,000 dictionaries items and they have no performance problems If the dictionary is sized correctly, there generally is no performance hit. However, editing it with some tools is a pain because it takes quite a long time to read them. Regards, Doug www.u2logic.com XLr8Dictionary Editor for large dictionary editing On Tue, Aug 6, 2013 at 9:08 PM, jeffrey Butera jbut...@hampshire.eduwrote: I'm curious how large of a DICTionary some of you have worked with and, in particular, how very large DICTs can adversely affect applications. We have a DICT approaching 1500 data elements (no idescs) - which is quite large for us. But I'm curious if others have DICTs this large or larger and have no adverse affect on their application performance. This is Unidata 7.3.4 if it matters. -- Jeffrey Butera, PhD Associate Director for Application and Web Services Information Technology Hampshire College 413-559-5556 __**_ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/**mailman/listinfo/u2-usershttp://listserver.u2ug.org/mailman/listinfo/u2-users ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users
Re: [U2] Large DICT affecting I/O
I would suggest that rather than the DICTIONARY being a performance problem, which can be easily overcome with a suitably sized file, the REAL impact may be in day to day processing where this file might be getting referenced to display a dozen or so fields, but each I/O is actually pulling in much, Much, MUCH more, pulling in 2 orders of magnitude more fields than it needs to Sure, UD/UV/MV ALLOWS you to DO this sort of thing, but I would respectfully suggest that you shouldn't necessarily do so. If people adopt bad habits like this, they tend to just keep on evolving, and I suspect if performance is currently OK (thanks to ever faster hardware), there is probably no desire to go through the pain associated with fixing this (or budget :-) -Original Message- From: u2-users-boun...@listserver.u2ug.org [mailto:u2-users-boun...@listserver.u2ug.org] On Behalf Of jeffrey Butera Sent: Wednesday, August 7, 2013 1:09 PM To: U2 Users List Subject: [U2] Large DICT affecting I/O I'm curious how large of a DICTionary some of you have worked with and, in particular, how very large DICTs can adversely affect applications. We have a DICT approaching 1500 data elements (no idescs) - which is quite large for us. But I'm curious if others have DICTs this large or larger and have no adverse affect on their application performance. This is Unidata 7.3.4 if it matters. -- Jeffrey Butera, PhD Associate Director for Application and Web Services Information Technology Hampshire College 413-559-5556 ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users ___ U2-Users mailing list U2-Users@listserver.u2ug.org http://listserver.u2ug.org/mailman/listinfo/u2-users