Thanks to Ruks Omar blog Big Data webinar - Strategies for implementing your Big Data using an Architecture Approach you can now watch a recording of vijay.vijayasankar Big Data - Avoiding Big Data Chaos For Customers And Developers webcast this past week.
I took some notes during the webcast, and they are below. Please note they are rough notes but some interesting concepts.
Figure 1: Source: SAP
Vijay said that big data is not a separate solution.
Big data is just data, just more of it in different ways
Figure 2: Source: SAP
Figure 2 shows IT's perfect definition of big data. Vijay said you can’t express 3V model to business user though.
The business shouldn’t have to know volume.
The only V that matters to business is value and that comes from analysis of that data
Figure 3: Source: SAP
Sheer number of tools is unbelievable.
Tools make it hard to understand
"Tools don’t define a strategy" said Vijay.
Tools assemble data in a way that are easy to use, and you need synergy to use them
With a Big Data Platform, the user needs to be shielded from the tools
Open source community has different tools and technologies; too many of them with overlap
BI tools that will do a little of everything or need 5 of them to make sense
The car analogy was used - the car has a dashboard to show you where you are going, speed, but not need to know codes for what is causing trouble in engine
"Shield complexity from the user"
Figure 4: Source: SAP
Value comes from the analytics but also putting it into action
Using tactical terms, use this to read the data you want
Action could mean writing back to the platform
If you are trying to answer a call in call center and have big data solution at disposal, customer discounts, and cross-sell – only makes sense if you can act on it
Using both decision and context - the analogy he used was you go and buy a house – budget is $300K, one month – Google search – 3 houses at $300K – information is useful if you know the context of the house – school system, airport, high crime.
We have always had this data, but missed the context around the decision
Another example was sales. It is a bad idea to send Collections over when you have a bigger deal going on
Not everything needs processed at real time - there is the idea of choice
XS in HANA:
- App server allows building of applications
- As you cut down the number of layers it becomes less tedious as the data is next to the logic
- Doesn’t mean use another way of architecting a solution – JDBC, ODBC
Figure 5: Source: SAP
Deep analysis is one thing, as shown in Figure 5
- Go to the lowest level of information
- High speed – real time versus right time – if something is in batch, lose opportunity to act on it – like the stock market – split second decision making and actions, more questions, drill in
- Simple – otherwise no adoption
SQL is not going anywhere
Before HANA could not optimize across all 5 dimensions
We may be moving to a time where "speed is taken for granted"
Figure 6: Source: SAP
Smart Data Access helps “minimize chaos” (shown in Figure 6)
Text analytics has natural language processing with automatic language detection
Spatial offers native geo-spatial types – HANA one, BW on HANA – tactical use cases – some companies re-do regions. In classical data warehousing need a hierarchy with geospatial and now it can immediately understand it
Application development is easier and more sophisticated
Check HANA roadmap in SMP; also on saphana.com
Figure 7: Source: SAP
Keep core data where you want
SP7 has other ways and an SDK to have federation with other sources
You don’t need to transfer data from old to new
You can do analytics on HANA and keep data in old systems and HANA will talk to old systems to get the data you want – HANA developer doesn’t need to know how to access in Teradata – can access as if in HANA
Developer is shielded from underlying chaos
Figure 8: Source: SAP
Figure 8 shows the key take-aways from the webcast
Question and Answer:
Q: How often get new things in HANA?
A: Revisions come every few weeks, at least once a month
Every few months get new SP’s
Q: Is HANA platform big data platform SAP only?
A: any developer should be able to build on it; refer Bangalore TechEd keynote
MOOCs for free training
Intention is to create a large ecosystem for HANA
Q: What deployment options if choose HANA platform?
A: do not force a deployment option
HANA Cloud Platform – SAP’s HANA as a service offering
Develop on XS natively
- SAP Managed Tags:
