I'm trying to create a datamart for the healthcare application. The facts in the datamart are basically going to be measurements and findings related to heart, and we have 100s of them. Starting from 1000 and can go to as big as 20000 per exam type.
I'm wondering what my design choices for the fact tables are:
Grain: 1 row per patient per exam type.
Some of the choices that I can think of -
1) A big wide fact table with 1000 or more columns.
2) EAV based design - A separate Measure dimension table. This foreign key will go into the fact table and the measure value will be in fact table. So the grain of the fact table will be changed to 1 row per patient per exam type per measurement.
3) Create smaller multiple fact tables per exam type per some other criteria like subgroup. But the end user is going to query across subgroups for that exam type and fact-fact join is not recommended.
4) any other ideas?
Any inputs would be appreciated.
1. A big wide fact table with 1000 or more columns.
One very wide fact table gives end-user maximum flexibility if queries are executed directly in the data warehouse. However some considerations should be taken into account, as you might hit some limits depending on a platform.
SQL Server 2014 limits are as per below:
Bytes per row 8,060. A row-overflow storage might be a solution, however it supports only few column types typically not related to fact nature, i.e. varchar, nvarchar, varbinary, sql_variant. Also not supported in In-Memory OLTP. https://technet.microsoft.com/en-us/library/ms186981(v=sql.105).aspx
Columns per non-wide table 1024. Wide-tables and sparse columns are solution as columns per wide table limit is 30,000. However the same Bytes per row limit applies. https://technet.microsoft.com/en-us/library/cc280604(v=sql.120).aspx
Columns per SELECT/INSERT/UPDATE statement 4,096
Non-clustered indexes per table 999
https://technet.microsoft.com/en-us/library/ms143432(v=sql.120).aspx
2. EAV based design - A separate Measure dimension table. This foreign key will go into the fact table and the measure value will be in fact table. So the grain of the fact table will be changed to 1 row per patient per exam type per measurement.
According to Kimball, EAV design is called Fact Normalization. It may make sense when a number of measurements is extremely lengthy, but sparsely populated for a given fact and no computations are made between facts.
Because facts are normalized therefore:
Extensibility is very easy, i.e. it's easy to add new measurements without the need to amend the data structure.
It's good to extract all measurements for one exam and present measurements as rows on the screen.
It's hard to extract/aggregate/make computation between several measurements (e.g. average HDL to CHOL ration) and present measurements/aggregates/computations as columns, i.e. requires complex WHERE/PIVOTING or multi-joins. SQL makes it difficult to make computations between facts in different rows.
If primary end-user platform is an OLAP cube then Fact Normalization makes sense. The cubes allows to make computation across any dimension.
Data importing could be an issue if data format is in a flat style CSV.
This questions is also discussed here Should I use EAV model?.
3) Create smaller multiple fact tables per exam type per some other criteria like subgroup. But the end user is going to query across subgroups for that exam type and fact-fact join is not recommended.
In some scenarios multiple smaller fact tables perfectly makes sense. One of the reason is if you hit some physical limits set by platform, e.g. Bytes per row.
The facts could be grouped either by subject area, e.g. measurement group/subgroup, or by frequency of usage. Each table could be placed on a separate file group and drive to maximize I/O.
Further, you could duplicate measurements across different fact tables to reduce the need of fact tables join, i.e. put one measurement in a specific measurement subgroup fact table and in frequently used measurement fact table.
However some considerations should be taken into account if there are some specific requirements for data loading. For example, if a record errors out in your ETL to one fact table, you might want to make sure that the corresponding records in the other fact tables are deleted and staged to your error table so you don't end up with any bogus information. This is especially true if end users have their own calculations in the front end tool.
If you use OLAP cubes then multiple fact tables actually becomes a source of a measure group to a specific fact table.
In terms of fact-to-fact join, you (BI application) should never issue SQL that joins two fact tables together across the fact table’s foreign keys. Instead, the technique of Drilling Across two fact tables should be used, where the answer sets from two or more fact tables are separately created, and the results sort-merged on the common row header attribute values to produce the correct result.
More on this topic: http://www.kimballgroup.com/2003/04/the-soul-of-the-data-warehouse-part-two-drilling-across/
4) any other ideas?
SQL XML or some kind NoSQL could be an option, but the same querying / aggregation / computation / presentation issues exist.
Related
We are implementing a new dwh solution. I have many dimensions that require slowly changing type 2 attributes. I was considering implementing a combination of Type 2 and Type 1 attributes in my dimension. That is for some dimension attributes, we track history by inserting new rows in the dim table (Type2), for other attributes we will just update the existing row for any changes (Type1)
Questions:
Is this a good practice? is it OK to have a combination of SCD 1 and 2 for the same dim?
Is there any limit on the number of SCD 2 attributes in a dimension? My dimension is pretty wide, like 300 cols, out of which one of the users is requesting for about 150 cols to be tracked by scd type 2. Is it OK to have so many scd2 attributes in a dim? Is there going to be any impact on performance of downstream reporting BI solutions like cubes and dashboards because of this?
In the OLTP system, we maintain an "audit" table to log any updates. Though this is not in a very easily queryable format, we get answers to most of our questions related to changes from this. We don't need much reporting on data changes. Of course there are some important columns like Status for which we definitely need SCD2 but rest of the columns, I am not sure having history for lot of other columns in the DWH adds any value. My question is when we have this audit table in OLTP, how do I decide what attributes need SCD 2 in the DWH?
Good practice? Yes. Standard feature of dimensional modelling that is overlooked too often. I've seen dimensions with combinations of SCD0, SCD1 and SCD2, and there's nothing to prevent other SCD-types being used as well.
No limit on columns, but that does seem a little excessive. You probably want to use a "hash" method to detect the SCD2 changes, where you calculate a hash over the SCD2 columns, and use this value to detect if any of the columns have changed.
Sorry, but I don't understand the question about audit logs. Are these logs your data source?
I'm investigating data warehouses. And I have an issue about star schemas.
It's in
Oracle® OLAP Application Developer's Guide
10g Release 1 (10.1)
3.2.1 Dimension Table: TIME_DIM
https://docs.oracle.com/cd/B13789_01/olap.101/b10333/global.htm#CHDCGABE
To represent the hierarchy MONTH -> QUARTER -> YEAR, we need some keys such as: YEAR_ID, QUARTER_ID. But there are some things that I do not understand:
1) Why do we need field YEAR_DSC & QUARTER_DSC? I think that we can look up these values from YEAR & QUARTER TABLE. And it breaks 2NF.
2) What is the normal form that a schema in data warehouse needs to satisfy? (1NF, 2NF, 3NF, or any.)
NFs (normal forms) don't matter for data warehouse base tables.
We normalize to reduce certain kinds of redundancy so that when we update a database we don't have to say the same thing in multiple places and so that we can't accidentally erroneously not say the same thing where it would need to be said in multiple places. That is not a problem in query results because we are not updating them. The same is true for a data warehouse's base tables. (Which are also just queries on its original database's base tables.)
Data warehouses are usually optimized for reading speed, and that usually means some denormalization compared to the original database to avoid recomputation at the expense of space. (Notice though that sometimes rereading something bigger can be slower than reading smaller parts and recomputing the big thing.) We probably don't want to drop normalized tables when moving to a data warehouse, because they answer simple queries and we don't want to slow down by recomputing them. Other than those tradeoffs, there's no reason not to denormalize. Some particular warehouse design methods might have their own rules about what parts should be denormalized what amounts.
(Whatever our original database design NF is chosen to be, we should always first normalize to 5NF then consciously denormalize. We don't need to normalize or know constraints to update or query a database.)
Read some textbook basics on why we normalize & why we use data warehouses.
I'm trying to design a Data Warehouse for a single store of commonly required data ranging from finance systems, project scheduling systems and a myriad of scientific systems. I.e. many different data marts.
I have been reading up on Data Warehousing and popular methods such as Star Schemas and Kimball methods etc but one question I cannot find answer to is:
Why is it better to design your DW Data Mart as a star schema rather than a single flat table?
Surely having no joins between facts and attributes/dimensions is faster and simpler than having lots of small joins to all the dimension tables? Disk space is not a problem, we'll just throw more disks at the database if necessary. Is the star schema slightly outdated these days or is it still data architect dogma?
Your question is very good: the Kimball mantra for dimensional modelling is to improve performance and to improve usability.
But I don't think it is outdated, or dogma- it is a reasonable, practical approach for many situations and platforms.
The way relational DBs store data means there's a balancing act to be struck between the numbers and types of tables, the routes in to the data for typical queries, easy maintainability and description of relationships between data, the numbers of joins, the way the joins are constructed, the indexability of columns, etc.
3NF (or further) is one end of the spectrum, suiting OLTP systems, and a single table is the other end of the spectrum. Dimensional models are in the middle and appropriate for reporting, at least when using certain technologies.
Performance isn't all about 'number of joins', although a star schema performs better for reporting workloads than a fully normalised database, in part because of a reduce number of joins. Dimensions are typically very wide. If you are including all those dimension fields in every row of every fact, you have very large rows indeed, and finding your way into those rows will perform very badly for typical queries.
Facts are numerous, so if you can make those tables compact, with the 'wordier' dimensions filterable, you hit a sweet spot of performance that a single table isn't going to match, unless heavily indexed.
And yes a single table for a fact is simpler in terms of numbers of tables but is it really easier to navigate? Dimensions and facts are easy concepts to understand, and what if you want to cross you queries across facts? You've got many different data marts but one of the benefits of having a data warehouse in the first place is that these aren't distinct- they're related and can be reported across. Conformed dimensions enable this.
If you combine your fact and dimensions into a single table, you'll either lose the visibility into dimension attributes that have never been used, or your measures will be thrown off by inclusion of a dummy event for the unused dimension attribute.
For example, a restaurant menu is a dimension and the purchased food is a fact. If you combined these into one table, how would you identify which food has never been ordered? For that matter, prior to your first order, how would you identify what food was available on the menu?
The dimension represents the possibilities, the fact represents the realization of the possibilities.
Combining facts and dimensions in the same table limits the scalability and the flexibility.
Suppose that one day the business decides to change a dimension description ( for example the product name ). Dimension tables aren't as deep as the fact tables and the update process or SCD management should be easier and less resource intensive.
Imagine the following situation I am planning:
Have two rather large tables stored in Hive, both containing different types of customer related information (say, although this is not exactly the case, a record of customer transactions in one and customer owned data in the other). Let's call the tables A and B.
Tables are large in the sense that none of the tables fits completely in memory. (There are 10 million customers and theres is a few kilobytes of info associated to each of them in each of the two tables)
Be careful enough to bucket both tables in exactly the same way, by a field present in both tables (customer_id, which is a bigint), and using the same number of buckets 100.
I wonder whether this setup will, in any way, guarantee that a join (by customer_id) between both tables will be efficient, in the sense that very little shuffling of information between nodes will be required. I imagine this could the case, if for instance, there were a guarantee that the physical files corresponding to the same bucket in both tables are physically stored in the same (sets of nodes), i.e. if for every bucket i (in [0,99]) the file A/part_0_000i and the file B/part_0_000i were physically stored in the same nodes and the same held for their replicas.
Notes:
I am aware that partitioning and bucketing are different and that the first essentially determines the structure of subdirectories, whereas the second on determines which file each record goes too. This question is about bucketing only
Also, by number 2, map-side joins are not an option here, since, as far as my understading goes, they require loading one of the tables completely within each mapper and doing the join completely there.
Bucketing is used when there are too many levels in your data in which you want to partition by, or there are no good candidate partitions.
A concrete example would be partitioning on customerID in sales data. You may have 20 thousand customers. Partitions would contain small amounts of data which is inefficient and have too many partitions also inefficient. However you can hash the customerID and partition into 50 buckets for example. Then when you are merging on customerID the job will only have to scan against what is in a bucket rather than the entire sum of all your data.
With ideal bucketing your buckets should contain some multiple of the file system block size. Remember also that too many buckets or buckets that are built over varialbes not used as keys can be detrimental for other queries.
I have used them when I need to execute large jobs repeatedly. My queries time has been reduced significantly. I tend to only use when my data is very big. And big is relative to cluster size and capacity.
One great thing about bucketing is that they help ensure the bucketed partitions are of similar size. If you partition over State for example, California will have huge partitions while other states are very small.
Bucketing is tactical and not an appropriate for all use cases. Happy bucketing!
Yes, it will definitely help.
Bucketed tables are partitioned and sorted the same way, so they will be mergesorted, which works in linear time (n), otherwise the tables have to be sorted the same way first, which is usually nlog(n)
I've a bunch of columns in my fact tables that have a very low cardinality (~8). Each of these columns store keys that refer to a master table. I'm wondering whether to import each of these individual master tables as dimension or do I store the values directly in the fact table. Master tables have no additional attributes except the value I'm trying to store. What are the pros and cons of each approach ?
This seems to be a classic example of a junk dimension that combines together a number of miscellaneous, low-cardinality flags and indicators (instead of putting each of them in a separate dimension table).
Disadvantages of other approaches:
Putting every low cardinality attribute in a separate, dedicated dimension could result in an overly complex model with excessive number of dimension tables (centipede fact tables).
Storing the attributes directly in the fact table is allowed but reserved only for degenerate dimensions, i.e. values like order or invoice numbers, retail point-of-sale transaction numbers - high-cardinality values that don't have any additional attributes describing them.
Low-cardinality flags are not DDs, because even though they may consist of a sole key now, they may easily have additional attributes in the future, e.g. multiple descriptive captions for reports - short for mobile users and long for desktop users.
Details: Design Tip #113 Creating, Using, and Maintaining Junk Dimensions