ppt

Report
PLANNING FOR
CONVERGED
INFRASTRUCTURE
Performance, Capacity and Management
Peter Dyer – Product Manager
www.uptimesoftware.com
Why am I here?
• Alex couldn’t be here
• Building software for the datacenter for more than 15
years
• Focused on virtualization for the past 8 years
• Worked on a tool called PlateSpin Recon – one of the first
virtualization consolidation/capacity planning tools
• Working with up.time customers to define the future of
our product, including addressing management and
capacity planning challenges in the modern datacenter
www.uptimesoftware.com
Definition –
Converged Infrastructure
Grouping of multiple components: server (compute), storage,
networking and management software in a single, optimized
computing package
+
+
=
or
www.uptimesoftware.com
Two approaches
Integrated Systems
Fabic Based (Hyperconverged)
(older, scale up)
www.uptimesoftware.com
(newer, scale out)
www.uptimesoftware.com
Subcategories (Gartner) –
Integrated Systems (Converged Infrastructure)
• Integrated stack systems (ISS) – server, storage and
network with application software
– Simplified: really big server (one application)
• Integrated infrastructure systems (IIS) – server, storage
and network to provide compute infrastructure
– Simplified: private cloud
• Fabric-based computing (FBC) – server, storage and
network together in an integrated system designed to be
building blocks
– Simplified: private cloud building blocks
www.uptimesoftware.com
Integrated stack systems (ISS)
Examples: Oracle Exadata, IBM PureApplication, Teradata
www.uptimesoftware.com
Integrated infrastructure systems (IIS)
Examples: VCE Vblock (VMware, Cisco, EMC), HP
ConvergedSystem, IBM PureFlex, Cisco-NetApp FlexPod
www.uptimesoftware.com
Fabric-based computing (FBC) /
Hyperconverged Infrastructure
Examples: SimpliVity, Nutanix, HP Moonshot
www.uptimesoftware.com
Demand and Supply
Integrated Systems
www.uptimesoftware.com
Fabric-based
www.uptimesoftware.com
Under the covers
• Integrated systems
–
–
–
–
Existing technology
Reference architecture
Pre-configured
Tested
• Fabric-based
– New technology
– Existing parts
www.uptimesoftware.com
The History of Fabric-based
Converged Infrastructure
Google
www.uptimesoftware.com
Key use cases
• Private cloud
• Virtual desktops
• Applications
– Traditional data driven: Oracle, SAP, etc
– Big data: Hadoop, etc
• Greenfield
www.uptimesoftware.com
Business Drivers
•
•
•
•
•
Improved performance
Lower operating costs
Simplified support
Faster deployment
Simplified sourcing
www.uptimesoftware.com
Market
• Gartner estimates
– Converged infrastructure (Integrated systems)
• 2014: $6B
• 2014: 50% growth rate
– Total hardware market: $80B
• MarketsandMarkets
– Converged Infrastructure
• $33.89B by 2019
www.uptimesoftware.com
Systems Management
• Good news
– Workloads are not changing
• Operating systems and applications just see (virtual) hardware
– Hypervisor is not changing
• But there are more (new) parts
• Bad news
– Hardware is changing, and access to physical hardware is changing
www.uptimesoftware.com
Not isolated
Converged infrastructure
Private cloud
www.uptimesoftware.com
Software-defined
data center
Not isolated
Converged infrastructure
Private cloud
www.uptimesoftware.com
Software-defined
data center
Software-Defined Data Center
Storage
• VMware
– vSANs
– vVols
• Microsoft
– Storage spaces
• EMC
– ScaleIO
Networking
• VMware
– NSX
• Microsoft
– Hyper-V Network
Virtualization (HNV)
• Cisco
– Extensible Network
Controller (XNC)
• VMware
– EVO (Rail & Rack)
www.uptimesoftware.com
www.uptimesoftware.com
Software-Defined Storage (SDS)
• Pool of disks treated as one disk resource
– Local disk
• VMware vSAN
• Microsoft Storage Spaces
• Fabric-based (Hyperconverged) solutions
– External disk (NAS, JBOD, etc)
• VMware vVols
• Goals
– Lower costs
– Easier to scale
– Easy management (eliminates hardware RAID and proprietary SAN
operating systems)
www.uptimesoftware.com
Storage is evolving (sort of)
• Capacity is ever increasing, IOPS sort of
• IOPS is king (or at least very important)
– Mixing of different types of storage – fast tier
– Dynamic tiering
– Caching
• Replication
– Data protection
www.uptimesoftware.com
Software-Defined Networking (SDN)
• Moving switch management from the physical port to
software that follows the VM (workload)
• Goals
– Lower costs
– Easy management
– Better security (egg)
www.uptimesoftware.com
The Software-Defined Challenge
• Who knows the truth
– It used to be the operating system – it isn’t any more
www.uptimesoftware.com
The Old Days
Application
www.uptimesoftware.com
www.uptimesoftware.com
The Old Days (monitored)
Application
www.uptimesoftware.com
www.uptimesoftware.com
…along comes SAN
Application
www.uptimesoftware.com
www.uptimesoftware.com
…then comes virtualization
Application
VM
VM
VM
Hypervisor
www.uptimesoftware.com
www.uptimesoftware.com
VM
VM
…software-defined data center
SDN
Application
VM
VM
VM
Hypervisor
www.uptimesoftware.com
www.uptimesoftware.com
VM
VM
SDS
…Public Cloud
SDN
Application
VM
VM
VM
Hypervisor
www.uptimesoftware.com
www.uptimesoftware.com
VM
VM
VM
SDS
…monitoring
SDN
Application
VM
VM
VM
Hypervisor
www.uptimesoftware.com
www.uptimesoftware.com
VM
VM
VM
SDS
The New Way Tradeoff
• Good
– Time to provision new systems (agility) – decreased dramatically
– Cost per system – decreased dramatically
• Bad
– Cost to manage new systems – increased dramatically
– Number of systems – increased dramatically
www.uptimesoftware.com
Not isolated
Converged infrastructure
Private cloud
www.uptimesoftware.com
Software-defined
data center
Private Cloud Defined
• Shared pool of computing resources inside a corporate
firewall
• Characteristics
–
–
–
–
Easy, fast provisioning / access to resources
Used by many departments / groups
Used for a wide variety of use cases
Buzz words: agile, automated
www.uptimesoftware.com
Cloud Capacity Challenge
• Are people using the resources they have requested?
– Generally, no they are not
www.uptimesoftware.com
Strategies to address challenge
• Show costs upfront at time of provisioned
• Chargeback
– Show-back
• Shame-back
• Lease only (must be renewed)
• Overprovision
www.uptimesoftware.com
Not isolated
Converged infrastructure
Private cloud
www.uptimesoftware.com
Software-defined
data center
Capacity Planning Challenge
• Limited ability to isolate capacity and performance to
discrete elements
– CPU, memory, disk (fast and slow) purchased together
www.uptimesoftware.com
vBlock example
(small and simple)
www.uptimesoftware.com
www.uptimesoftware.com
Related challenge
• Fast and slow disk
www.uptimesoftware.com
vBlock example
(small and simple)
www.uptimesoftware.com
www.uptimesoftware.com
Nutanix
www.uptimesoftware.com
www.uptimesoftware.com
Simplivity Omnicube
www.uptimesoftware.com
www.uptimesoftware.com
WRAP UP
www.uptimesoftware.com
www.uptimesoftware.com
Questions
•
•
•
•
What type of converged infrastructure is being deployed?
What is it being used for?
How easy is it to add capacity?
How much capacity must be purchased?
– How much lead time is required?
• Can capacity be optimized across compute and storage?
• What is the maximum workload size that needs to be
supported?
www.uptimesoftware.com
Takeaways
• Converged Infrastructure doesn’t exist in isolation –
– Software-defined data center
– Use case (ie Private Cloud, VDI, etc)
• Software-defined data center introduces “source of truth”
management challenge
• Private Cloud introduces capacity challenges
• Converged Infrastructure introduces capacity challenges
www.uptimesoftware.com
Map Dashboard
Capacity Dashboard
Topology Dashboard
Thank you.
[email protected]
Global Scan
www.uptimesoftware.com
Resource Scan

similar documents