bigdata

@theCUBE Talks Big Data
Conversation with George Gilbert Wikibon Analyst about bigdata Hadoop & other cool news & trends
   5 years ago
#BigDataSystems of IntelligenceConversation with Wikibon #bigdata research analyst on Systems of Intelligence & it's impact
   4 years ago
#BigDataBecome a Big Data ChampionFrom data lakes to data analytics – gain advantage over your competition. Chat with the coaches!
John Furrier
What can we learn from other mega trends in the past and how that relates to bigdata #hadoop etc
George Gilbert
in the mid- to late-90's #ERP, internet infrastructure sw revenue grew triple digits
George Gilbert
but they all crashed into a wall starting in '98 because customers had bought much faster than they could deploy - some in anticipation of #Y2K
George Gilbert
there is a similar sense today of "fear of missing out" driving #bigdata purchases while the technology is still maturing
George Gilbert
that's what our survey data and research in the #bigdata ecosystem tells us
George Gilbert
for example - between spring '14 and fall '15, those describing their deployments as "successful" has grown from 41% then to 45% now.
John Furrier
So do you see the Hadoop market crashing @hortonworks is releasing earnings today
George Gilbert
part of the push for triple digit but unsustainable growth in the late '90s came from the vendors themselves
George Gilbert
enterprise #software could cost well over 50% of every $ in revenue because of cost of paying expensive sales reps
George Gilbert
the problem of the cost of sales is worse today because #opensource #software doesn't come with the huge upfront licenses that came with software in the '90s
John Furrier
I think hadoop is still messy but the foundational elements are in place now to extend to other systems that should propel growth; what do you think?
George Gilbert
so vendors have an even greater motivation to sell more to the customer than they can absorb in any near term period
George Gilbert
#hadoop isn't a product but rather an ecosystem. parts of that like #yarn and #hdfs are ubiquitous and will be a foundation as the #analytics stack grows and matures
George Gilbert
for customers to get past complexity, we are likely to see convergence of more components
George Gilbert
for example, most #hadoop vendors are placing #apachespark as the core of their #analytics frameworks while keeping the more mature management and admin bits from hadoop
John Furrier
How do public cloud platforms figure into plans?
George Gilbert
public cloud is going to big the big wildcard. most customers and vendors regard them as direct competitors to #hadoop right now
George Gilbert
but in our survey an equal % used #hadoop and native #aws #azure #googlecloudplatform services - both with roughly 37%
George Gilbert
this question specifically asked about the use of public #cloud for #bigdata #analytics
George Gilbert
the reasons for use of public #cloud were 58% data already there and 54% for operational simplicity #TCO
George Gilbert
44.5% said public #cloud gave developers #analytics tools that were easier to build an end-to-end application with versus on-premises
John Furrier
AWS is crushing it bc they have the app marketplace; IBM is right behind them building fast
Brian Gracely
if customers can get the data to the public clouds (either natively or thru transfer), the path to get started and maintain the systems is much simpler. tons of working big data examples on public clouds.
John Furrier
. @bgracely totally agree but those apps are not #hadoop but other underlying tech #integratedstacks
Brian Gracely
AWS EMR is Hadoop.
John Furrier
@bgracely is EMR considered an app?
Brian Gracely
EMR is foundational, whereas Kinesis or Redshift would be considered apps. either way, they are all simplifying (or operationalizing) the part customers have problems getting to work.
Brian Gracely
is CrowdChat a Hadoop or Big Data application?
John Furrier
@bgracely CrowdChat is a dynamoDB app and a big data app on AWS
Dormain Drewitz
@bgracely yes, and a key word is "if". To @ggilbert41's point, it makes a lot of sense if the data is already there
rob harrington
George - when do you think you start to see some of the potential stresses highlighted in your report play out in terms of the big data and hadoop pure play companies? how will this play out?
George Gilbert
the #bigdata revenue numbers we see today from individual vendors aren't as big as the late '90s, so we may have headroom but...
George Gilbert
when there is rampant and steep discounting in a market as part of a land grab, that's often an indicator that we're close to a correction
George Gilbert
in addition, there are way too many vendors - public and private - that are spending so much on sales & marketing in this land grab that they are consuming significant amounts of the cash on their balance sheets
George Gilbert
that provides a potential vendor-induced retrenchment if the capital markets no longer accept heavy operating losses without evidence of future profitability
George Gilbert
it's not going to be easy to find many canaries in the coal mine because most of the vendors are private...
rob harrington
many thanks this makes sense. i sometimes hear commentary that cloudera has deep pockets but its all become too complicated & the code base has become too messy. its not as simple & clean cut offering like it was a while back - does any of that ring true?
rob harrington
thanks for your commentary?
George Gilbert
but maybe #hortonworks which reports today can provide some insight into what's going on in the #bigdata ecosystem more broadly
George Gilbert
#cloudera started life with the objective of making #hadoop easier to manage and they've done a good job. but the #cloud competition may be most intense because those vendors live to simplify and have the infrastructure built-in to do it
John Furrier
Share info about your research survey. What was the survey methodology?
George Gilbert
we surveyed 300 practitioners in U.S. this fall and also spring 18 months ago
George Gilbert
we surveyed enterprises that had deployed or were evaluating a #bigdata #analytics project
George Gilbert
we defined #bigdata #analytics to include technologies and data that traditional #sql #databases couldn't address
John Furrier
what are customers saying? will hadoop blowup or grow?
George Gilbert
#hadoop isn't going away anytime soon - it has bright prospects
John Furrier
. @ggilbert41 can hadoop pure play vendors survive or do they have to do more?
George Gilbert
the problem with #bigdata software currently is that vendors are selling it and customers are buying it faster than they can possibly deploy
Brian Gracely
@ggilbert41 do vendors see that as a problem, because they can't monetize the apps/customizations yet, or just viewed as a customer problem?
George Gilbert
@bgracely great question: i think we are likely to see both sides recognizing this as a problem
George Gilbert
@bgracely customers are essentially piling up "inventory" of undeployed licenses and vendors are facing a "channel" that's filling up
George Gilbert
@bgracely the problem is that customers may start wanting to deploy more in the #cloud for simplified admin and overall TCO
George Gilbert
@bgracely in addition - they may see the cloud native services such as #aws #kinesis #redshift #machinelearning as more accessible to developers and even simpler to operate
John Furrier
. @bgracely to me there are not enough big data apps out there on hadoop imho that is the big issue
Brian Gracely
this is why you're seeing things like AWS QuickSite (https://aws.amazon.c...) get created because companies have existing "Data Scientists on Excel" that could use simpler applications and shareable tools/data.
George Gilbert
also #ibm #watson #expertstorybooks are "notebooks" like we're seeing spring up in the #apachespark community and lately #hadoop too
George Gilbert
these go a long way toward simplifying #datascience and application building. #ibm calls them citizen #datascientists