This presentation was given at the London Nutanix user group (NUG) on Oct 26 by Ray Hassan. If you would like to join a NUG, you can find more information here http://bit.ly/NTNXUG - Hope to see you at a community meeting!
Next-generation AAM aircraft unveiled by Supernal, S-A2
Got Big Data? Splunk on Nutanix
1. Big Data : Splunk on Nutanix
Ray Hassan
ray@nutanix.com
@cannybag
2. 2
Nutanix in 30 Seconds: Invisible
Infrastructure
Just
Works
Eliminates
Guesswork
Removes
Constraints
Invisible
Infrastructure
Why What
Hyperconvergence
How
Web-Scale
3. 3
Nutanix Web-Scale Architecture
Eliminates
SAN and
NAS
arrays
Tier 1 Workloads
(running on all nodes)
Nutanix Controller VM
(one per node)
Node 2
VM VM VM CVM
X86
Node N
VM VM VM CVM
X86
Node 1
VM VM VM CVM
X86
Local + Remote
(Flash + HDD)
Distributed Storage Fabric
intelligent tiering, VM-centric management and more…
ü Snapshots ü Clones ü Compression ü Deduplication
ESX
i
Acropolis App Mobility Fabric
AH
V
Hyp
er-V
ESX
i
AH
V
Hyp
er-V
ESX
i
AH
V
Hyp
er-V
Workload
Mobility and
Hypervisor
Choice
ü Locality ü Tiering ü DR ü Resilience
4. 4
The Digital Universe….
"100% of all large enterprises
will adopt [Operational
Intelligence] technologies for
big data analytics within the
next two years,“ Forrester
On average, between 60% and 73%
of all data within an enterprise goes
unused for business intelligence (BI)
and analytics, Forrester
Source: Forrester’s Global Business Technographics® Data And Analytics Survey, 2015
44ZB
DATA
------------------
2020
….doubling in size every
two years, IDC
5. 5
Key benefits of Splunk on Nutanix
Ability to ingest GBs of data
per day
Quick search capabilities for
mission critical applications
Ability to support growth in
data ingest rates
1 TB+/day of data ingest
ample for most deployments
Accelerated search through
server side flash
Predictable, linear performance
through distributed architecture
Self contained deployments
due to data security & privacy
Quick, manageable
deployment through appliance
10. 10
NUTANIX Cluster
Node 1 Node 2
Indexer Cluster
Peer 1 Peer 2
Splunk>
CVM
Node N
CVM CVM
RF2
Master node
No Master failover within Splunk
Indexer Peer Node
Cluster/Bucket Fixing - buckets either made searchable/streamed to
other peers or held in reserve for eventual return of node:
• very CPU intensive
• impact network interconnect
High Availability - Protect the Master
11. 11
Site 1 Site 2
VM 1 … VM n VM 1 … VM n
RF2 RF2
site_replication_factor = origin:1, site1:1, site2:1, total:2
NUTANIX
Splunk>
site_replication_factor = origin:2, site1:1, site2:1, total:3
NDFS : Redundant data path
12. 12
Tested Platform (SplunkIT)
High Performance
Index Rate > 500K EPS
Avg. TFE ~ 1.9 secs
Avg. TTS ~ 16 secs
56K IOPS/2U
Technical Partners
Reference
Architectures, Tech
Field Day Demo,
Blogs
Sales Engagements
Nordstroms,
NASDAQ, Covance,
Labcorp, Nintendo etc.
13. 13
Save on Archiving and Licensing
NX-8150 Series
Compute and storage
NX-6035C Series
Storage only
10GbpsEthernet
IOPS Storage
14. 14
Automatic Disk Balancing
ü Real-time balancing of storage within the cluster/nodes
ü Supports heterogeneous and homogenous node types (Compute heavy/Storage heavy)
ü Uniform distribution of data
ü Leverages MapReduce framework
ü Requires no manual intervention
This process is done both during runtime with node/disk placement as well as
back ground Curator process
Larger aka “Storage Heavy” nodes will have larger capacities
and hence hold more data
After disk balancing has run the utilization
will be uniform
NDFS
Hypervisor Hypervisor Hypervisor
VM 1 VM N VM 1 VM N VM 1 VM N
Storage Storage Storage
CVM CVM CVM35% 35%
35%
17. 17
3 GB/s Sequential
100,000 Random Read IOPS
500,000 EPS
2U Cluster
Nutanix/Splunk
RA available on
Nutanix.com
Splunk on Nutanix Reference Architecture
18. 18
Nutanix Solutions for Enterprises
VDI
Branch
Office
Data Protection
& Disaster
Recovery
Enterprise
Applications
Private &
Hybrid Clouds
Microsoft
Applications
Big Data