HYPERCONVERGED INFRASTRUCTURE Adam Savage A Brief history of
HYPERCONVERGED INFRASTRUCTURE Adam Savage
A Brief history of the data centre • The Mainframe • • Natively converged CPU, memory and storage Had redundancy built-in Proprietary & Expensive Lack of flexibility and scalability
Evolution of the datacenter Application Node 1 Node 2 Node N C 1 C 2
Evolution of the datacenter VM VM VM Hypervisor Node 1 Node 2 Node N New C 1 Flash C 2 Flash
Complexity Every Step of the Way DEPLOY Deployment takes weeks to months -- complex, big footprint SUPPORT Multi-hop support, lack of end-to-end visibility, firefighting BUY MANAGE Purchase in big increments, incur prediction risk, overprovision Multiple management panes, manual operations GROW Difficult to scale quickly
Evolution of the datacenter VM VM VM CVM VM VM Hypervisor Node 1 Node 2 Node N Single Storage Pool Local (Flash + HDD) Natively Supports Snapshots Clones Compression De-duplication CVM
Scale Out Architecture NDFS Pooled resources NDFS 10 Gbps Ethernet SINGLE NODE Compute: 16 Cores Storage: Flash: 800 GB SSD HDD: 4 TB Controller: Nutanix Storage controller NUTANIX INC. – CONFIDENTIAL AND PROPRIETARY
Why Nutanix?
How does v. Motion/DRS work? VM data ABC VM data DEF VM data GHI Nutanix Controller VM Hypervisor AD B G C D A I B E CF A G H DC
How does v. Sphere High Availability (HA) work? data VM VM VM ABC data ABC DEF VM data GHI Nutanix Controller VM Hypervisor AD B G C D A I B E CF A G H DC
Why Nutanix? • Don’t rely on network performance for reads or initial writes (just redundancy copying) • RAID is a thing of the past – Imagine we striped and mirrored all of our hosts via RAID 0+1 – Rely on network to read and write data – Now try to grow your environment
Our Deployment
Our Deployment
Our Deployment
- Slides: 18