Ultra-high I/O ECSs

Overview

Ultra-high I/O ECSs use high-performance local NVMe SSDs to provide high storage input/output operations per second (IOPS) and low read/write latency. You can create such ECSs with high-performance local NVMe SSDs attached on the management console.

**Table 1** Ultra-high I/O ECS features
Series	Compute	Disk Type	Network
I3nl	vCPU to memory ratio: 1:8 Number of vCPUs: 8 to 64 2nd Generation Intel® Xeon® Scalable Processor Basic/Turbo frequency: 3.0 GHz/3.5 GHz	Ultra-high I/O General Purpose SSD High I/O Extreme SSD	Ultra-high PPS throughput An ECS with higher specifications has better network performance. Maximum PPS: 8,000,000 Maximum intranet bandwidth: 40 Gbit/s
I3	vCPU to memory ratio: 1:8 Number of vCPUs: 8 to 64 Intel® Xeon® Scalable Processor Basic/Turbo frequency: 3.0 GHz/3.4 GHz	Ultra-high I/O General Purpose SSD High I/O Extreme SSD	Ultra-high PPS throughput An ECS with higher specifications has better network performance. Maximum PPS: 5,000,000 Maximum intranet bandwidth: 25 Gbit/s
I7n	vCPU to memory ratio: 1:4 Number of vCPUs: 8 to 96 3rd Generation Intel® Xeon® Scalable Processor Basic/Turbo frequency: 2.6 GHz/3.5 GHz	Ultra-high I/O General Purpose SSD High I/O Extreme SSD	Ultra-high PPS throughput An ECS with higher specifications has better network performance. Maximum PPS: 8,000,000 Maximum intranet bandwidth: 40 Gbit/s

Scenarios

Ultra-high I/O ECSs are suitable for high-performance relational databases.
Ultra-high I/O ECSs are suitable for NoSQL databases (such as Cassandra and MongoDB) and ElasticSearch.

Specifications

**Table 2** I7n ECS specifications
Flavor	vCPUs	Memory (GiB)	Max./Assured Bandwidth (Gbit/s)	Max. PPS (10,000)	Max. NIC Queues	Max. NICs	Max. Supplementary NICs	Local Disk	Virtualization
i7n.2xlarge.4	8	32	10/3.4	120	4	4	64	1 × 1.6T NVMe	KVM
i7n.8xlarge.4	32	128	25/13.5	400	8	8	192	4 × 1.6T NVMe	KVM
i7n.12xlarge.4	48	192	30/20	500	16	8	256	6 × 1.6T NVMe	KVM
i7n.16xlarge.4	64	256	35/27	600	16	8	256	8 × 1.6T NVMe	KVM
i7n.24xlarge.4	96	420	44/40	800	32	8	256	12 × 1.6T NVMe	KVM

**Table 3** I3 ECS specifications
Flavor	vCPUs	Memory (GiB)	Max./Assured Bandwidth (Gbit/s)	Max. PPS (10,000)	Max. NIC Queues	Max. NICs	Local Disks	Virtualization
i3.2xlarge.8	8	64	2.5/2.5	100	4	4	1 × 1,600 GiB NVMe	KVM
i3.4xlarge.8	16	128	5/5	150	4	8	2 × 1,600 GiB NVMe	KVM
i3.8xlarge.8	32	256	10/10	200	8	8	4 × 1,600 GiB NVMe	KVM
i3.12xlarge.8	48	384	15/15	240	8	8	6 × 1,600 GiB NVMe	KVM
i3.16xlarge.8	64	512	25/25	500	16	8	8 × 1,600 GiB NVMe	KVM

**Table 4** I3nl ECS specifications
Flavor	vCPUs	Memory (GiB)	Max./Assured Bandwidth (Gbit/s)	Max. PPS (10,000)	Max. NIC Queues	Max. NICs	Local Disks	Virtualization
i3nl.2xlarge.8	8	64	15/4	120	4	4	1 x 1.6 TiB NVMe	KVM
i3nl.4xlarge.8	16	128	20/8	224	8	8	2 x 1.6 TiB NVMe	KVM
i3nl.8xlarge.8	32	256	30/16	440	16	8	4 x 1.6 TiB NVMe	KVM
i3nl.12xlarge.8	48	384	35/27	750	16	8	6 x 1.6 TiB NVMe	KVM
i3nl.16xlarge.8	64	512	40/32	800	32	8	8 x 1.6 TiB NVMe	KVM

Local Disk Performance

Table 5 and Table 6 list the IOPS performance of local disks and specifications of a single local disk attached to an I7n ECS.

**Table 5** IOPS performance of local disks used by I7n ECSs
Flavor	Maximum IOPS for Random 4 KB Read
i7n.2xlarge.4	900,000
i7n.8xlarge.4	3,600,000
i7n.12xlarge.4	5,400,000
i7n.16xlarge.4	7,200,000
i7n.24xlarge.4	10,800,000

**Table 6** Specifications of a single local disk attached to an I7n ECS
Metric	Performance
Disk capacity	1.6 TB
IOPS for random 4 KB read	900,000
IOPS for random 4 KB write	250,000
Read throughput	6.2 GiB/s
Write throughput	2.1 GiB/s
Access latency	Within microseconds

Table 7 and Table 8 list the IOPS performance of local disks and specifications of a single local disk attached to an I3 ECS.

**Table 7** IOPS performance of local disks used by I3 ECSs
Flavor	Maximum IOPS for Random 4 KB Read
i3.2xlarge.8	750,000
i3.4xlarge.8	1,500,000
i3.8xlarge.8	3,000,000
i3.12xlarge.8	4,500,000
i3.15xlarge.8	5,250,000
i3.16xlarge.8	6,000,000

**Table 8** Specifications of a single I3 local disk
Metric	Performance
Disk capacity	1.6 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

Table 9 and Table 10 list the IOPS performance of local disks and specifications of a single local disk attached to an I3nl ECS.

**Table 9** IOPS performance of local disks used by I3nl ECSs
Flavor	Maximum IOPS for Random 4 KB Read
i3nl.2xlarge.8	750,000
i3nl.4xlarge.8	1,500,000
i3nl.8xlarge.8	3,000,000
i3nl.12xlarge.8	4,500,000
i3nl.16xlarge.8	6,000,000

**Table 10** Specifications of a single local disk attached to an I3nl ECS
Metric	Performance
Disk capacity	1.6 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

If the host where an ultra-high I/O ECS is deployed is faulty, the ECS cannot be restored through live migration.
- If the host is faulty or subhealthy, you need to stop the ECS for hardware repair.
- In case of system maintenance or hardware faults, the ECS will be redeployed (to ensure HA) and cold migrated to another host. The local disk data of the ECS will not be retained.
Ultra-high I/O ECSs do not support specifications change.
Ultra-high I/O ECSs do not support local disk snapshots or backups.
Ultra-high I/O ECSs can use local disks, and can also have EVS disks attached to provide a larger storage size. Note the following when using the two types of storage media:
- Only an EVS disk, not a local disk, can be used as the system disk of an ultra-high I/O ECS.
- Both EVS disks and local disks can be used as data disks of an ultra-high I/O ECS.
- An ultra-high I/O ECS can have a maximum of 60 attached disks (including VBD, SCSI, and local disks).
Modify the fstab file to set automatic disk mounting at ECS start. For details, see Configuring Automatic Mounting at System Start.
The local disk data of an 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 an 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, an 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 an ultra-high I/O ECS is /dev/nvme0n1 or /dev/nvme0n2.