Katie & Emil >BI Tutorials > SSIS interview questions and ...

1 Katie & Emil >BI Tutorials > SSIS interview questions and answers SSIS interview questions and answers If you need to interview someone with SSIS skills and you do not know much about the tool itself (or you are just after some examples) than it is worth knowing what are good SSIS interview questions and answers to ask. If you are on the other side and soon you will be attending an interview then this post might help in your preparations. The questions and answers below are for SQL Server Integration Services 2008 R2 but I will also explain briefly differences for 2008 and 2005. Currently I will only post several questions and I will add new ones when I have a minute. Below are categorized SSIS interview questions and answers : General SSIS questions question : Which versions of SSIS have you used? Answer: Differences between 2005 and 2008 are not very big so 2005, 2008 or 2008 R2 experience usually is very similar.

2 The big difference is with 2000 which had DTS and it very different (SSIS is created from scratch) question : Have you used SSIS Framework? Answer: This is common term in SSIS world which just means that you have templates that are set up to perform routine tasks like logging, error handling etc. Yes answer would usually indicate experienced person, no answer is still fine if your project is not very mission critical. question : Do you have experienced working with data warehouses? Answer: SSIS is in most cases used for data warehouses so knowledge of Data Warehouses Designs is very useful. question : How have you learnt SSIS (on the job, articles, books, conferences) Answer: The thing is that most people who read good books have usually an advantage over those who hasn't because they know what they know and they know what they don't know (but they know it exists and is available).

3 Blog/Articles very in quality so best practise articles is a big plus+, conferences can be also a plus. question : Do you have certifications Answer: This is rather disappointing point for me. Qualifications generally are welcome but unfortunately many people simply cheat. Companies run courses and then give questions and answers , or people find them on the internet. I've met people who had certification but knew very little, I've met people very experienced and knowledgeable without certification and people who have done certification for their self-satisfaction and are experienced and knowledgeable. In other words be careful with It is easy to get a misleading impression so make sure you ask the best questions for the position you can. question : What are the changes in SSIS 2012? Answer: Even if you don't use SSIS 2012 this questions checks if candidate is interested in SSIS.

4 Major changes are configuration,deployment and logging. question : If you have been given opportunity to improve SSIS. What would it be? Answer: I have been asked this question on an interview and I LOVE it! This is great question as it check how well someone knows the tool itself. The more valid comments the more knowledge this person has and because this questions is brilliant! I'm not going to reveal answers here ;) SSIS Development questions question : How many difference source and destinations have you used? Answer: It is very common to get all kinds of sources so the more the person worked with the better for you. Common ones are SQL Server, CSV/TXT, Flat Files, Excel, Access, Oracle, MySQL but also Salesforce, web data scrapping. question : What configuration options have you used? Answer: This is an important one. Configuration should always be dynamic and usually is done using XML and/or Environment Variable and SQL Table with all configurations.

5 question : How do you apply business rules in SSIS ( calculations but also cleansing)? Answer: Some people use SSIS only to extract data and then go with stored procedures are usually missing the point of the power of SSIS. Which allows to create "a flow" and on each step applies certain rules this greatly simplifies the ETL process and simplicity is very good. question : How to quickly load data into sql server table? Answer: Fast Load option. This option is not set by default so most developers know this answer as otherwise the load is very slow. question : Give example of handling data quality issues? Answer: Data Quality is almost always a problem and SSIS handles it very well. Examples include importing customers from different sources where customer name can be duplicates. For instance you can have as company name: SQL Server Business Intelligence but also SQL Server BI or SQL Server BI LTD or SQL Server BI Limited or intelligence (with one l).

6 There are different ways to handle it. Robust and time consuming is to create a table with or possible scenarios and update it after each update. You can also use fuzzy grouping which is usually easy to implement and will make usually very good decisions but it is not 100% accurate so this approach has to be justified. Other typical quality issues are nulls (missing values), outliers (dates like 2999 or types like 50000 instead of 5000 especially important if someone is adjusting the value to get bigger bonus), incorrect addresses and these are either corrected during ETL, ignored, re-directed for further manual updates or it fails the packages which for big processes is usually not practised. question : When to use Stored Procedures? Answer: This was one of the requested question in comment (at the bottom of the page). This one is very important but also tricky.

7 ALL SSIS developers have SQL Server background and that is sometime not very good if they use SQL not SSIS approach. Let's start with when you typically use SPs. This is for preparing tables (truncate), audit tasks (usually part of SSIS framework), getting configuration values for loops and a few other general tasks. During ETL extract you usually type simple SQL because it comes from other sources and usually over complication is not a good choice (make it dynamic) because any changes usually affect the package which has to be updated as well. During Transformation phase (business rules, cleaning, core work) you should use Transformation tasks not Stored procedures! There are loads of tasks that make the package much easier to develop but also a very important reason is readability which is very important for other people who need to change the package and obviously it reduces risks of making errors.

8 Performance is usually very good with SSIS as it is memory/flow based approach. So when to use Stored Procedures for transformations? If you don't have strong SSIS developers or you have performance reasons to do it. In some cases SPs can be much faster (usually it only applies to very large datasets). Most important is have reasons which approach is better for the situation. question : What is your approach for ETL with data warehouses (how many packages you developer during typical load etc.)? Answer: This is rather generic question . A typical approach (for me) when building ETL is to. Have a package to extract data per source with extract specific transformations (lookups, business rules, cleaning) and loads data into staging table. Then a package do a simple merge from staging to data warehouse (Stored Procedure) or a package that takes data from staging and performs extra work before loading to data warehouse.

9 I prefer the first one and due to this approach I occasionally consider having extract stage (as well as stage phase) which gives me more flexibility with transformation (per source) and makes it simpler to follow (not everything in one go). So to summarize you usually have package per source and one package per data warehouse table destination. There are might be other approach valid as well so ask for reasons. SSIS Advanced questions question : What is XMLify component? Answer: It is 3rd party free component used rather frequently to output errors into XML field which saves development time. question : What command line tools do you use with SSIS ? Answer: dtutil (deployment), dtexec (execution), dtexecui (generation of execution code) PDF Download - ebook I will make this page available in PDF format for download (eBook) soon. I will try to find time to add more questions soon.

10 Feel free to suggest new questions (add comments). Hope that helps! Emil See also: SSRS interview questions and answers