SLA и гарантии надёжности хостинга Windows

Main / Cloud services catalog / Cloud Server / SLA и гарантии надёжности хостинга Windows

Cloud Server (IaaS)

Виртуальные серверы на базе Windows Server 2016 Hyper-V и AzurePack

Крупнейший в России ЦОД TIER-III Facility & Operation
Оборудование производства HP, Huawei, CISCO
До 72 ядер Intel Xeon Scalable Gold 3,7 ГГц
До 1,5 ТБ памяти на каждый виртуальный сервер
Финансовая гарантия SLA — 99,9 %
Бесплатный еженедельный BackUp

категории

SLA and Windows Hosting Reliability Guarantees

Financially guaranteed SLA — 99.9 %

If the Windows Azure service level specified in the agreement is not met, the customer shall receive compensation calculated as a percentage of the monthly service fee applicable at the time of the incident:

Total availability time (x)	Total downtime during a calendar month	Compensation as a percentage of the monthly service fee
x ≥ 99.9%	Up to 45 minutes	0%
99.0% ≤ x < 99.9%	From 45 minutes to 7.5 hours	25%
95.0% ≤ x < 99.0%	From 7.5 hours to 36 hours	50%
x < 95.0%	More than 36 hours	100%

High Availability of Virtual Servers

All Azure cloud virtual machines run on a fault-tolerant Hyper-V cluster. If any physical server fails, the virtual machines are automatically restarted on standby servers. All physical data storage systems are equipped with at least two fault-tolerant controllers, and all disks are combined into fault-tolerant RAID arrays. In addition, dedicated virtual servers are backed up weekly to dedicated backup storage, with two backup copies retained. This makes the risk of data loss virtually negligible.

Contractually Guaranteed Performance

No.	Computing resource	Minimum guaranteed performance	Measurement procedure
1.	vCPU_H (Intel Xeon Scalable Gold 4.0 GHz)	4000 MIPS (The guaranteed frequency of all cores under 100% load is at least 3.9 GHz)	Performance is measured using the 7-Zip archiver in a single-threaded benchmark using the total rating. All application software must be closed during measurement. The test must run for at least 5 minutes.
2.	vCPU_M (Intel Xeon Scalable Gold 3.0 GHz)	3000 MIPS (The guaranteed frequency of all cores under 100% load is at least 2.9 GHz)
3.	vCPU_L (Intel Xeon Scalable Gold 2.4 GHz)	2000 MIPS (The guaranteed frequency of all cores under 100% load is at least 2.2 GHz)
4.	vRAM	100% reserved capacity (dynamic memory is not used) Guaranteed memory speed: 6 GB/s	Memory speed is measured using the Memory Speed Test x64 utility, Fill Memory test, 100 MB
5.	vHDD SSD NVME OPTANE	Sequential speed: 2500 MB/s Random performance: 500,000 IOPS Backup: once a week, 2 restore points	Testing is performed using CrystalDiskMark software. IOPS is measured in multithreaded mode (8–64 threads) using 4K blocks. Tests must be performed outside the agreed backup schedule because performance may be reduced during backup operations.
6.	vHDD SSD ULTRA	Sequential speed: 1000 MB/s Random performance: 100,000 IOPS Backup: once a week, 2 restore points
7.	vHDD SSD HIGH	Sequential speed: 500 MB/s Random performance: 50,000 IOPS Backup: once a week, 2 restore points
8.	vHDD SSD MEDIUM	Sequential speed: 500 MB/s Random performance: 5,000 IOPS Backup: once a week, 2 restore points
9.	vHDD SAS_HIGH	Sequential speed: 300 MB/s Random performance: 1,000 IOPS Backup: once a week, 2 restore points
10.	vHDD SAS_LOW	Sequential speed: 200 MB/s Random performance: 500 IOPS Backup: once a week, 2 restore points
11.	vHDD SATA	SLA is not guaranteed (average performance: approximately 100 MB/s) Backup: not provided
12.	vLAN	Connection speed: 10 Gbps Throughput: approximately 10 Gbps Minimum guaranteed throughput: 1 Gbps Average latency: approximately 1 ms Maximum guaranteed latency: 5 ms Packet loss: no more than 0.1%	Bandwidth can be measured using JPerf. Latency can be measured using the Ping utility. The specified parameters must be measured over a 10-minute period, during which no other applications may generate network traffic.
13.	WAN 100M	Connection speed: 100 Mbps Throughput: approximately 100 Mbps Minimum guaranteed throughput: 50 Mbps Average latency: approximately 5 ms Maximum guaranteed latency: 20 ms Packet loss: no more than 1%	The Internet connection speed can be measured using Speedtest.net or a similar service. No other applications may generate Internet traffic during testing. The results provided by such services are not absolute in any case, and service quality must be assessed based on a combination of factors and multiple tests.

All performance metrics specified in the table are measured by the Customer from within the virtual machine. If unsatisfactory results are obtained, the Customer must contact the Contractor’s technical support service and provide the corresponding measurement results. The Contractor may request access to the operating system of the Customer’s virtual server in order to verify the test results, and the Customer must provide such access.
If any performance metric specified in the table falls below 50% of its stated value, the corresponding computing resource shall be deemed completely unavailable, and its monthly fee shall be compensated in accordance with the SLA. In all other cases, the Customer may request a proportionate reduction in the cost of the affected resources based on the duration of the SLA violation and the measured percentage of performance degradation.
The performance violation period shall be the time interval between the moment the performance degradation is confirmed by the Contractor’s engineers and the moment the incident is resolved.

24/7 Support

Round-the-clock technical support is provided with a guaranteed response time of 15 minutes.

Fault-Tolerant Server and Network Architecture

Four fibre-optic links to the core CISCO switches of the data centre network.
A stack of Huawei 10G Ethernet switches with 2N redundancy.
Virtualisation servers combined into a Hyper-V cluster.
A SAN fabric consisting of Huawei 8G Fibre Channel switches with 2N redundancy.
A pool of dual-controller hybrid storage systems using SSD, SAS and NL-SAS drives.
A pool of iSCSI/SMB storage systems with SATA drives for backup storage.