As data continues to explode across the enterprise, we are finding that it is becoming increasingly challenging for organizations to keep up. A recent Splunk report, "The Data Age is Here," found that 57% of companies interviewed expressed that the volume of data is growing faster than they can manage, with 47% bluntly saying they will fall behind when faced with rapid data volume growth. The good news is that this is the perfect opportunity to leverage streaming technology, and why we are excited to announce the general availability of Splunk Data Stream Processor (DSP) 1.2!
So, what’s in the box?!?
Gain Control Over Multicloud Sprawl
DSP 1.2 provides users with the ability to expand data access and support for multicloud environments where data sprawl creeps in. We now include GCP Pub/Sub sources to support our customers multicloud strategy and the need to move data between multiple vendors. We’ve also added Azure Event hub to the growing list of supported destinations.
We all know IT data can be more valuable if enriched with context however, adding this context on the stream to high volume data with millisecond latency is no easy task. DSP 1.2 Lookups allow you to do in-stream data enrichment at scale, unlocking additional value in IT and Security. DSP Lookups help enrich events in your data pipeline with contextual and timely data, making downstream searches in Splunk Enterprise much more relevant and accurate. From a security perspective, this can help improve overall security posture by increasing the fidelity of incident detection and accelerating investigations for determining “who” or “what” was involved in, or affected by an incident.
Yes – machine learning! Streaming ML unleashes your ability to learn, infer and analyze data. Unlike traditional batch ML systems where you have to train and validate models and constantly update them, our ML learns continuously so your models are always up to date. Moreover, our machine learning scales seamlessly, independent of volume and cardinality. Our first-of-their-kind algorithms on the stream that can learn with a “single pass” of data will accelerate your insights, and DSP 1.2 is adding to our growing list of algorithms, including:
- Online time series decomposition: or online STL, normalizes data sets by decoupling seasonality. This accelerates pattern detection in things like inventory management which can lead to greater accuracy in purchasing, stocking and staffing decisions
- Categorical outlier detection: useful in security or IT domains when trying to detect outliers with non-numeric data values
- Approximate percentiles: helping you categorize data in specific groupings so you can answer questions like which machines are generating the top traffic, or cellphones that are in the top 5% of bandwidth consumption
We also make applying ML to the stream easy with a dedicated ML GUI so folks who aren’t data scientists can capture the value of machine learning.
For more detail on how you can apply machine learning to the stream, check out our latest essential guide here.