avg time for reports bulding

[sempaj]

Dear All,

nowadays we are going to deploy Cognos BI, May you help with clarification of wordays for implementation of new reports on Cognos BI base:
      a)   usage of exiting data marts & DWH data;
      b)   creation of new data marts & existing DWH data;
      c)   usage of existing data marts & upload addition data from core system to DWH;
      d)   creation of new data marts & upload addition data from core system to DWH;

kind regards,
Alexandr

[cognostechie]

There is no such thing as an average time to build either reports or anything else. It depends on the level of effort required to build anything depending on what you have available, how many different source systems or external data you have, how much effort is required to create Data Marts/DW , models cubes etc.

The estimation is done based on the extent of source data involved. Based on that we can estimate the time needed to build a proper Data Mart or Data Warehouse and then modelling. If the model/cubes are proper then it takes less time to build reports.

[sempaj]

many thanks for prompt answer. Also please look at clarification: all data are placed in one DWH. the DWH is built on IBM DWH model.
Could you re-send your estimation?

[sempaj]

any idea?

[blom0344]

sempaj,

You are asking too much  ;)   There are clearly too many aspects involved.

I also wonder why such questions are asked in the first place. An experienced Cognos developer would - given enough time - be able to give you an idea when he/she could look into required report designs and models to be developed.

There is no way you can get a relevant answer to these questions on a technical forum

[sempaj]

i try to create a calculator for prompt time estimation of report creation(target ~300 reports).
taking into account my experience i will apply following figures:
      a)   usage of exiting data marts & DWH data - one work day in avg: report development can need from 15 mon to ~1 day;
      b)   creation of new data marts & existing DWH data- 2 workdays in avg: 1 day for report & 1 for data mart;
      c)   usage of existing data marts & upload addition data from core system to DWH - 3 workdays in avg: 1 report, 0.5 data mart update, 1.5 day for data upload upgrade(in case of data existing in core systems);
      d)   creation of new data marts & upload addition data from core system to DWH - 4 work days: 1 for report+ 1 for data mart ~2 for the upload;
please adjust/comment the figure & approach
i try to create a calculator for prompt time estimation of report creation(target ~300 reports).
taking into account my experience i will apply following figures:
      a)   usage of exiting data marts & DWH data - one work day in avg: report development can need from 15 mon to ~1 day;
      b)   creation of new data marts & existing DWH data- 2 workdays in avg: 1 day for report & 1 for data mart;
      c)   usage of existing data marts & upload addition data from core system to DWH - 3 workdays in avg: 1 report, 0.5 data mart update, 1.5 day for data upload upgrade(in case of data existing in core systems);
      d)   creation of new data marts & upload addition data from core system to DWH - 4 work days: 1 for report+ 1 for data mart ~2 for the upload;

please adjust/comment the figures & approach

[CognosPaul]

While it may occasionally be acceptable to throw out numbers like that, you absolutely must consider the complexity of the project, requests, and any future requirements.

The very first thing you can do, before you can even start estimating number of reports or even time it may take to develop the datamarts, is to interview everyone. Talk to the stakeholders, talk to the end users, talk to the people who run the system, talk to the DBAs. Once you get everything down, schedule another round of meetings to go over what you've learned from everyone else.

I really cannot stress enough the important of the requirement gathering stage. A truly successful project absolutely needs the support of the end users and stakeholders, and it needs clearly stated goals.

1. What do the end users expect to get?
2. What do the stake holders expect to achieve?
3. What can the ETL guys provide?
4. What database limitations are set by the DBAs?
5. At what level of granularity is your data? Does it meet the end user needs? Can you easily expand it to provide more atomic information if necessary?
6. What do the stake holders expect the end users to need? How do they react to the end users stated needs?
7. Ask the users again, with the information you've learned, are there any other things the end user needs? How will the end users use the reports? Do the end users anticipate needing to use data exploration or adhoc querying tools?
8. Talk to the Cognos administrator. Discuss the end user requirements. Does your organization have enough licenses to cover their needs? Will the Cognos server support the additional load? Do you have international users who need dedicated gateways or dispatchers?

The requirement gathering stage alone shouldn't take less than a week. Do not rush it. Write down everything. Sketch out reports with the end user. Discuss possible alternatives. Sketch out possible models with the DBA. Talk to the ETL guys about any foreseeable issues increasing complexity.

By the time this stage is done, you should have a much better idea of how long it would take to develop your reports.

As an example, I have just finished helped a client finish a series of reports that have been under development for a month. Due to the complexity of the requirement, we were actually allocated a bit more than that.

I have another client that was able to put together a script that churned out generic framework models as needed. Their clients would pay for specific features, and the generator would spit out the model.xml to meet their license.

The point I'm trying to make is that any attempt to estimate the amount of time needed, without first doing due diligence will only result in drastically inaccurate estimates. If I have a DWH of 3 tables, one fact and two dimensions, I could probably build a model in under an hour, and a several of reports by the EOB if I was rushing. But without seeing the structure of those three tables, I won't even commit to a day for the model.

It sounds like you're considering pulling from a DWH into one or more datamarts. Why not go the extra step and start constructing a series of cubes?

[sempaj]

thanks for your answer,
may you define the time according to your experience?

[cognostechie]

Can somebody pl. tell me the average time to become a millionaire ?  ;)

[CognosPaul]

That's exactly the point. Any numbers I give you will match the scenario of my projects, not necessarily yours.

The two extremes:

Project A: Ongoing for the past 4 years. The model itself was completely redesigned about 3 times in total and is constantly being updated as the DWH gains new data sources. The reports themselves have scheduled reviews and an average lifespan of 6 months (except for a standard suite of reports). The project is ongoing and is constantly evolving to meet the constant changes that the clients need. This is 4 years of constant development from myself and another developer covering for me when I'm not there.

Project B: A client that wanted an extremely simple model and set of reports. 4 tables, 3 distinct reports. Because everything was spelled out from the beginning (every single field was described, joins were defined, reports were diagrammed, sample outputs were provided based on existing data in the DB, expected response times were set) the actual development time was one day.

So based on my experience the range is one day with everything already spelled out and no room for misunderstandings to 4+ years for a constantly evolving set of needs. It would be incredibly unprofessional of me to give a basic "I estimate n months for this project" without having a significantly better understanding of what's involved. As you have the information, you will have to use and trust your own judgement to get at the numbers.

[blom0344]

Paul, I am a little disappointed that you do not unveal your palantir to give more exact estimates.  The OP surely isn't asking for much , is he?   ???

[blom0344]

Can somebody pl. tell me the average time to become a millionaire ?  ;)

14 years 3 months 11 days on average.  Unless you post a lot on Cognoise, then it takes a little longer   ;D

[sempaj]

so, i will use my estimation only,
additional question: how much time you need to make a coffee?

[cognostechie]

14 years 3 months 11 days on average.  Unless you post a lot on Cognoise, then it takes a little longer    ;D

So I will take less time than you !  ;D