Today we had a Tech Talk at work that was focused on our organizations upgrade to the newly released Splunk 6, and all its bells and whistles. This presentation was given by James Brodsky, a Sr. Sales Engineer from Splunk.
There are 3 main usage feature changes that have occured in splunk 6. The main focus of these upgrades are focused on increasing "non-technical" user experiences.
- Data Model
Also rather than reading through the below notes, you can also go through a great tutorial that introduces these new features.
The idea behind pivot is to utilize data models and graphical options to generate advanced queries. When using a pivot the user only needs to select the data model to use to view the pivotal data, and from there they can utilize simple combo boxes and links to create specific queries to use.
There is also an option to use Pivots as a starting point, and once you have filtered to the data you would like to see, you can convert the pivot to a search allowing you to use more advanced operations. This is achived using the pivot search command. It is also common to start the search with a | that has no start search, allowing for an "implicit search"
For a much more indepth look at pivots, please view the Splunk 6 documentation.
Data Models are fantastic, they are simply a meaningful representation of the underlying raw machine data. They are especially useful in allowing users to be consuming exactly the same data in their searches, rather than having each one define their own search attempting to gather the data (which will often be different).
To start creation on a Data Model you first need to understand how the model is setup. The model is based on constraints against the data. This allows for Data Models to be hierarichal by building the constraints on top of the previous constraint. This idea of constraints is completed by basically appending narrowing search fields.
index=production sourcetype=tomcat jvm=7
Because these constraints are based on the splunk search language, you can also use lookup tables when creating the data model, or other auto-extracted fields like sums that can be calculated rather than found in the data.
The key tie in for Data Models is their availability to use the new acceleration feature that can return data upto 1000x faster.
To help to speed up searches that use these data models, splunk has implemented Analytics Store. Note: This will increase storage and processing cost. The way this store works, is that it creates a custom index (basically) that will hold the data that is consumed by the data model. It is a sliding window store, so there is an initial increase in used storage when created, but after that should be relatively stable in and out (as long as the incoming longs are pretty set).
So here are the main benefits of an Analytics Store
- Benefit from Data Model requests for speed (up to 1000x faster)
- Allows for time sections for the increased analytics
There have been some new items that are integrated with Splunk 6 to increase its analytics.
- Integrated GEO-IP map that display geopraphic data
- MaxMind is the GEO backend licensed through Splunk
- Predict command on time series search
- Includes system capacity and resource utilization
- Faster, More Accurate and Flexible
Rich Developer Environment
- Dev Site
- Rapidly build pslunk apps using web technologies
- Allow for token passing easily in dashboards.
Changes in App Usage
As well as all these added features, there are some other features targeting just standard use of splunk.
- Drill down now has multiple options - None (Yeah), Inner, Outer and Full
- Search page additions - Inspect Search, Background Run, Progress Bar
- Apps show up on Home Page (Drag Drop capable)
- Time Picker broader selection
Splunk also now has a Splunk Hadoop Connect
Well after a great show of the new features, we now get to show the newly deprecated items, which is always best to keep till last. Luckily for this deprecation there is really only one thing that was brought up.
- Advanced XML Feature for Dashboards