Leaflet

© 2016 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information. Page 12
Cisco HyperFlex Systems
March 2016
there is no performance penalty for these operations. A small deduplication block
size helps increase the deduplication rate. Compression further reduces the data
footprint. Data is then moved to HDD storage as cache segments become free
(Figure 4).
Read operations cache data in the SSD drives and in main memory for high
performance. Having the most frequently used data stored in the caching layer
helps make Cisco HyperFlex Systems perform well for virtualized applications.
When virtual machines modify data, the original block is likely read from the cache,
so there is often no need to read and then expand the data on a spinning disk.
The data plaform decouples the caching tier from the capacity tier and allows for
independent scaling of I/O performance and storage capacity.
. . .
Node 1
Node n
Physical Layer
CPU and Memory NetworkHDDSSD CPU and MemoryNetwork HDD SSD
HDD-Based Capacity Layer
SSD-Based Caching Layer
. . . .
ESXi Hypervisor
VM VMVM
Data Platform
Controller
IO Visor
VAAI
VM_DIRECT_PATH
I/O Acceleration
ESXi Hypervisor
VM VMVM
Data Platform
Controller
VAAI
IO Visor
Replicate
Write to
Log on
SSD
Deduplicate
Compress
Destage
K
J
C
B
AD
F
E
C
B
A D
R T
V X
Z
ST
V X
Z
R S
A B B C C D
B
B
B
C
C
C
A B C D
A B B C C D
B
B
B
C
C
C
A B C D
VMware vSphere Plugin for Single-Pane
Hardware and Hyperconvergence Management
Cisco HyperFlex Administration Plugin
Cisco UCS Management Plugin
Figure 4. Data Write Flow Through the Cisco HyperFlex HX Data Platform