We have been talking much about our new Cloud server with SSD and its performance. Today, we want to make a comparison and benchmark on the cloud servers with spinning drives SSDs.
Vastspace SSD Cloud Server nodes use only enterprise SSD drives ensuring fast and consistent command response times as well as protect data loss and corruption.
We have done the read & write tests for our Cloud SSD VPS against a popular SSD VPS before. Today, we are carrying out test on 2 identical Cloud servers with SSD and Raid 10 15,000 rpm SAS drives respectively.
The test Cloud Servers comes with 2 CPU core, 2Gb memory and 20Gb of disk space.
Both test servers are installed with CentOS 6.5 x64 and hosted in Vastspace Singapore Data Center.
The result is obvious that SSD Cloud server beat the Cloud server with spinning drives hands down, despite the Raid 10 15K rpm SAS drives is still slower in terms of write speed compares to the solid state drives.
In general, many do not enquire the type of RAM used on their server. What most really care is the “number”, like I’ve 32Gb RAM and blah blah blah…
Do you know there are different type of RAM used on desktop computers, entry servers and high performance servers? It is important to find out that your server are fitted with ECC or better RAM for stability, reliability as well as performance for your 247 online activities.
Most branded entry dedicated servers use ECC RAM. ECC is an extension to parity uses multiple parity bits assigned to larger chunks of data to not only detect single bit error, but correct them automatically at the same time. Instead of the single parity bit for every 8 bits of data, ECC uses a 7 bit code that is automatically generated for every 64 bits of data stored in the RAM. When the 64 bits of data is read by the system, a second 7 bit code is generated, then compared to the original 7 bit code. If the codes match, the data is free of errors. If the codes don’t match, the system will be able to determine where the error is and fix them by comparing the two 7 bit codes.
Registered memory has registers or buffers included on the module for better flow of data which increases data reliability. It also allows for greater memory scalability and larger amounts of RAM can be installed. Because of this, registered memory is used mostly in servers with ECC functionality.
Fully buffered memory or also known as fully registered memory takes some of the functions from the memory controller. The communication between the memory controller and the module is serial, thus less number of wires is needed to connect the chipset to the RAM. With serial communication, fully buffered RAM is possible to have up to eight modules per channel and up to six memory channels, this greatly increasing RAM performance as well as memory scalability. Fully buffered memory cannot be used on a server that takes registered memory or vice verse. Fully buffered memory includes ECC functionality usually seen on high performance workstation and server.
When was the last time you have a visual check on your server internally? Some might have heard about the capacitor plague but if you have not, and you have owned the server for many years already you should do a visual inspection on your server motherboard immediately. The capacitor flaw was reported as early as 2002, and a surge of such complaints in 2010 with the higher than expected premature failure rate of aluminum electrolytic capacitors with the non-solid or liquid electrolyte capacitors from some Taiwanese manufacturers. The capacitors failed because due to a poorly formulated electrolyte with water-based corrosion effect.
Direct visual inspection is the most common method of identifying these capacitors which have failed because of bad electrolyte. Failed capacitors may show one or more of these visible symptoms:
a. Bulging or cracking of the vent on top of the capacitor.
b. Capacitor casing sitting crooked on the circuit board, as the bottom rubber plug has pushed out.
c. Electrolyte leaked onto the motherboard from the base of the capacitor or vented from the top, visible as crusty rust-like brown in colour deposits.
e. Detached or missing capacitor casing. Sometimes these vents do not open, a failed capacitor will literally explode, ejecting its contents violently and shooting the casing off the motherboard board.
When this happens, the capacitors no longer able to serve their purpose on filtering the direct current voltages on the motherboard, as a result of this failure is an increase in the ripple voltage that the capacitors are supposed to filter out. This causes system instability. Capacitors with high ESR and low capacitance can make power supplies malfunction, and causing further circuit damage. On the server, CPU core voltage or other system voltages may fluctuate and possibly with an increase in CPU temperature as the core voltage rises.
Even though there are not many cases nowadays and seems to have receded since 2013, I still urge you have this inspection done quickly especially for servers were manufactured before 2010. Apparently, I’m still seeing many of these branded servers are prone to this plague in operation when I was walking around the data centers.
Not all Cloud SSD VPS are the same, Vastspace SSD VPS server nodes are custom built and optimized RAID for redundancy and high performance. Vastspace SSD VPS server nodes use only Enterprise SSD Drives ensuring fast and consistent command response times as well as protect data loss and corruption.
We have done the read & write tests for our Cloud SSD VPS against a popular SSD VPS,
Our Cloud SSD VPS starts with 2 CPU core, 1Gb memory and 30Gb of disk space. For the tests, we have ordered the closer specification we can get from this provider, 2 CPU core, 2Gb memory and 40Gb of disk space.
Both candidates are based on CentOS 6.5 x64 and hosted in Singapore Data Centers
Here are the results;
Server Health monitoring is important and effectively provides useful additional perspective when combined with Up-time monitoring; it assists Vastspace in preventing downtime with informative capacity planning, rather than merely able to react to predictive failure or failure events. Common scenarios resulting in a server failure include excessive CPU usage, insufficient RAM and excessive Disk IO operations.
Our Server monitoring team is a dedicated team solely focused on providing service reliability and react to incident in proactive manner. The team utilizes an industry proven set of system level health and service monitoring tools to ensure our servers’ are optimal performance through early detection of problems. In the event that an issue is identified, our Team will respond timely, reducing downtime and fix any issues proactively, before the client is even aware of the problem at most time.