Updated on 2024-06-11 GMT+08:00

Kunpeng Ultra-high I/O ECSs

Overview

Kunpeng ultra-high I/O ECSs use Kunpeng 920 processors and 25GE high-speed intelligent NICs to provide up to 480 GiB DDR4-based memory with high network performance for large in-memory datasets.

Available now: kI1

Table 1 Kunpeng ultra-high I/O ECS features

Series

Compute

Disk Type

Network

kI1

  • vCPU to memory ratio: 1:4
  • Number of vCPUs: 8 to 64
  • Kunpeng 920 processor
  • Base frequency: 2.6 GHz
  • High I/O
  • General Purpose SSD
  • Ultra-high I/O
  • Extreme SSD
  • General Purpose SSD V2
  • Ultra-high packets per second (PPS) throughput
  • An ECS with higher specifications has better network performance.
  • Maximum PPS: 4,000,000
  • Maximum intranet bandwidth: 30 Gbit/s

Specifications

Table 2 kI1 ECS specifications

Flavor

vCPUs

Memory

(GiB)

Max./Assured Bandwidth

(Gbit/s)

Max. PPS

(10,000)

Max. NICs

Max. NIC Queues

Local Disks

Virtualization

ki1.2xlarge.4

8

32

7/3

80

4

4

1 × 3,200 GiB

KVM

ki1.4xlarge.4

16

64

12/6

140

6

4

2 × 3,200 GiB

KVM

ki1.6xlarge.4

24

96

15/8.5

200

6

8

3 × 3,200 GiB

KVM

ki1.8xlarge.4

32

128

18/10

260

6

8

4 × 3,200 GiB

KVM

ki1.12xlarge.4

48

192

25/16

350

6

16

6 × 3,200 GiB

KVM

ki1.16xlarge.4

64

228

30/20

400

6

16

8 × 3,200 GiB

KVM

Features

Table 3 and Table 4 lists the IOPS performance of kI1 ECSs and specifications of a single local disk attached to a kI1 ECS.

Table 3 kI1 ECS IOPS performance

Flavor

Maximum IOPS for Random 4 KB Read

ki1.2xlarge.4

750,000

ki1.4xlarge.4

1,500,000

ki1.6xlarge.4

2,250,000

ki1.8xlarge.4

3,000,000

ki1.12xlarge.4

4,500,000

ki1.16xlarge.4

6,000,000

Table 4 Specifications of a single NVMe disk attached to a kI1 ECS

Metric

Performance

Disk capacity

3.2 TB

IOPS for random 4 KB read

750,000

IOPS for random 4 KB write

200,000

Read throughput

2.9 GiB/s

Write throughput

1.9 GiB/s

Access latency

Within microseconds

Notes

  • For details about the OSs supported by Kunpeng ultra-high I/O ECSs, see OSs Supported by Different Types of ECSs.
  • Kunpeng ultra-high I/O ECSs do not support specifications modification.
  • Kunpeng ultra-high I/O ECSs do not support local disk snapshots or backups.
  • Kunpeng ultra-high I/O ECSs can use both local disks and EVS disks to store data. In addition, they can have EVS disks attached to provide a larger storage size. Use restrictions on the two types of storage media are as follows:
    • Only an EVS disk, not a local disk, can be used as the system disk of a Kunpeng ultra-high I/O ECS.
    • Both EVS disks and local disks can be used as data disks of a Kunpeng ultra-high I/O ECS.
    • A Kunpeng ultra-high I/O ECS can have a maximum of 60 attached disks (including VBD, SCSI, and local disks). Among the 60 disks, the maximum number of SCSI disks is 30, and the maximum number of VBD disks is 22 (including the system disk).
    • It is a good practice to use World Wide Names (WWNs), but not drive letters, to perform operations on local disks to prevent drive letter drift (low probability) on Linux. Take local disk attachment as an example:

      If the local disk WWN is wwn-0x50014ee2b14249f6, run the mount /dev/disk/by-id/wwn-0x50014ee2b14249f6 command.

      How can I view the local disk WWN?

      1. Log in to the ECS.
      2. Run the following command:

        ll /dev/disk/by-id

  • The local disk data of a Kunpeng ultra-high I/O ECS if an exception occurs, such as physical server breakdown or local disk damage. If your application does not use the data reliability architecture, it is a good practice to use EVS disks to build your ECS.
  • When a Kunpeng ultra-high I/O ECS is deleted, the data on local NVMe SSDs will also be automatically deleted, which can take some time. As a result, a Kunpeng ultra-high I/O ECS takes a longer time than other ECSs to be deleted. Back up the data before deleting such an ECS.
  • The data reliability of local disks depends on the reliability of physical servers and hard disks, which are SPOF-prone. It is a good practice to use data redundancy mechanisms at the application layer to ensure data availability. Use EVS disks to store service data that needs to be stored for a long time.
  • The device name of a local disk attached to a Kunpeng ultra-high I/O ECS is /dev/nvme0n1 or /dev/nvme0n2.
  • The basic resources, including vCPUs, memory, and image of a Kunpeng ultra-high I/O ECS will continue to be billed after the ECS is stopped. To stop the ECS from being billed, delete it and its associated resources.

Scenarios

Kunpeng ultra-high I/O ECSs can be used for high-performance relational databases, NoSQL databases (such as Cassandra and MongoDB), and ElasticSearch.