Why is StarWind VSAN better than a SAN array?
Jest wiele powodów wyboru DataCore Starwind vSAN zamiast fizycznej sieci SAN z macierzą RAID:
- Performance: latency in arrays is higher due to accessing disks via network rather than locally, no caching.
- Redundancy: a SAN can be a single point of failure; if a SAN device fails, the entire environment fails.
- High availability: The SAN array with 2 controllers (quite expensive) operates in static active-active mode, unlike StarWind's which operates in true dynamic active-active mode.
- Scalability: Enterprise license guarantees support for unlimited capacity in HA mode, this license is not time-limited. This means you can increase/scale by adding disk shelves when all the bays/bays are already full, without additional software costs. In addition, it is usually necessary to replace SAN components after 3-5 years, while StarWind's license is perpetual.
SAN elimination.
• Zarządzanie 2 serwerami jest łatwiejsze, wydajność jest lepsza z dyskami dostępnymi lokalnie, nie wymagana jest dedykowana sieć dla pamięci masowej. Klient może skonsolidować wszystkie swoje zasoby i pamięć masową, tylko na kilku hostach. Zbuduje w ten sposób infrastrukturę HCI (HyperConverged Infrastucture) zamiast oddzielnych serwerów i pamięci masowej. Jest to bardziej elastyczne i efektywne kosztowo. Zakup SAN dla 3-10 TB pamięci masowej nie ma sensu. Nawet jeśli nie zostaną poniesione duże koszty, to i tak w przeciągu 3-5 lat trzeba będzie kupić nowe urządzenie. SAN z 10-20 TB ma sens, tylko dla specyficznych obciążeń (m.in. wydajne bazy danych). W przypadku 20+ TB można się zastanawiać nad wdrożeniem SAN.
• Klienci z pamięcią masową do 20 TB potrzebują tylko dedykowanej pamięci masowej do backup’u. Cały ich storage jest na tych samych węzłach co maszyny wirtualne (VM) np. od Sangfor i aplikacje.
Comparing SANs with 2 controllers and 1 set of drives
Comparing a SAN with 2 controllers and 1 set of disks with a typical StarWind configuration (2 or more nodes with dedicated storage) is not a comparison of equivalent configurations. It is more appropriate to compare 2 SANs with 2 StarWind nodes.
- A single SAN is a single point of failure: one set of disks is vulnerable to failure.
- Two controllers are a good solution, but some arrays have problems with firmware upgrades during operation. This situation means that two controllers may not be enough.
- Two SANs offering what StarWind VSAN has: active-active replication, fault tolerance, redundant DRAM for write cache purposes, RDMA protocol connections. When you add these functionalities, you will find that more expensive array solutions are required.
In summary, StarWind's 2-node system offers great fault tolerance, performance and functionality compared to a SAN with two controllers and one set of drives.
A SAN is more costly in the long run:
- It cannot be updated indefinitely. 2-3 years of updates are not a problem, but after 3-5 years vendors often stop publishing updates, thus forcing the customer to buy a new system.
- StarWind runs on regular x86 systems, which can be easily upgraded. The license is perpetual and can be transferred to new hardware if necessary. The market for used x86 systems is much larger than used SAN systems without technical support (support).
Proactive support and monitoring.
Cheap arrays do not offer such functionality.