OpenStack large cloud

OpenStack large cloud

The Large Cloud is an OpenStack-based reference architecture for MCP. It is designed to provide a generic public cloud user experience to the cloud tenants in terms of available virtual infrastructure capabilities and expectations.

The large reference architecture is designed to support up to 5000 virtual servers or 500 hypervisor hosts. In addition to the desirable number of hypervisors, 18 infrastructure physical servers are required for the control plane. This number includes 9 servers that host OpenStack virtualized control plane (VCP), 6 servers dedicated to the StackLight services, and 3 servers for the OpenContrail control plane.

The following diagram describes the distribution of VCP and other services throughout the infrastructure nodes.

../_images/d_os_large_cloud.png

The following table describes the hardware nodes in the CPI reference architecture, roles assigned to them, and the number of nodes of each type.

Physical server roles and quantities

Node type

Role name

Number of servers

Infrastructure nodes (VCP)

kvm

9

Infrastructure nodes (OpenContrail)

kvm

3

Monitoring nodes (StackLight LMA)

mon

3

Infrastructure nodes (StackLight LMA)

kvm

3

OpenStack compute nodes

cmp

200 - 500

Staging infrastructure nodes

kvm

18

Staging OpenStack compute nodes

cmp

2 - 5

The following table summarizes the VCP virtual machines mapped to physical servers.

Resource requirements per VCP role

Virtual server roles

Physical servers

# of instances

CPU vCores per instance

Memory (GB) per instance

Disk space (GB) per instance

ctl

kvm02 kvm03 kvm04 kvm05 kvm06

5

24

128

100

dbs

kvm04 kvm05 kvm06

3

24

64

1000

msg

kvm07 kvm08 kvm09

3

32

196

100

prx

kvm07 kvm08

2

8

32

100

mdb

kvm07 kvm08 kvm09

3

8

32

150

TOTAL

16

328

1580

4450

Resource requirements per DriveTrain role

Virtual server roles

Physical servers

# of instances

CPU vCores per instance

Memory (GB) per instance

Disk space (GB) per instance

cfg

kvm01

1

8

32

50

cid

kvm01 kvm02 kvm03

3

4

32

500

TOTAL

4

20

128

1550

Resource requirements per OpenContrail role

Virtual server roles

Physical servers

# of instances

CPU vCores per instance

Memory (GB) per instance

Disk space (GB) per instance

ntw

kvm10 kvm11 kvm12

3

16

64

100

nal

kvm10 kvm11 kvm12

3

24

128

2000

TOTAL

6

120

576

6300

Resource requirements per Ceph role

Virtual server roles

Physical servers

# of instances

CPU vCores per instance

Memory (GB) per instance

Disk space (GB) per instance

cmn

kvm01 kvm02 kvm03

3

16

32

100

rgw

kvm01 kvm02 kvm03

3

16

32

50

TOTAL

6

96

192

450

Resource requirements per StackLight role

Virtual server roles

Physical servers

# of instances

CPU vCores per instance

Memory (GB) per instance

Disk space (GB) per instance

mon

kvm16 kvm17 kvm18

3

24

256

1000 0

mtr

kvm13 kvm14 kvm15

3

16

196

3000 0

log

kvm13 kvm14 kvm15

3

16

64 1

5000 2

TOTAL

9

192

1548

27000

0(1,2)

The required disk space per instance depends on the Prometheus retention policy, which by default is 5 days for mon nodes and 180 days for mtr nodes.

1

The Elasticsearch heap size must not exceed 32 GB. For details, see Limiting memory usage. To limit the heap size, see MCP Operations Guide: Configure Elasticsearch.

2

The required disk space per instance depends on the Elasticsearch retention policy, which is 31 days by default.

Note

  • The prx VM should have an additional NIC for the Proxy network.

  • All other nodes should have two NICs for DHCP and Primary networks.

See also