The appearance of network-attached storage (NAS) enables many homes and small to medium businesses to adopt economic method of storage equipment comparing with storage area network (SAN). The NAS provides a frame of file management that the homes and businesses can experience the complete and secure application. According to the cases that the writer noted, numbers of enterprises also place a NAS in a box to work as a back-up server and set up another NAS collocation internally to have double security of the data.

There are several elements for enterprises to evaluate purchasing a NAS which will also be the future trends.

Storage Capacity
A 2-Bay NAS server equipped with 2 500GB HDs / 1TB HD is a common choice for a small-size business when 4-Bay, 5-Bay or 8-Bay servers are more suitable for larger businesses.

RAID Array Systems
A hard drive is a hardware component of limited life span. When a hard drive failure occurred, the data stored is not easy to get restored. At this moment, the hard drive array system plays an important role. Many small-size businesses use 2-Bay NAS with 2 hard drives which follow the RAID I model for the mirroring. When one hard drive is down, the other can back up to replace the original one when medium-size businesses usually go for RAID 0, RAID 1, RAID 5, RAID 5 + Spare Disk (RAID 6) and JBOD for more options.

Bandwidth
A NAS can be as good as a data storage computer differentiated by CPU and RAM. The spec is determined by the processing time and bandwidth of the NAS. The bandwidth of an entry level NAS with Giga Lan can reach 10MB per minute as high-end models can reach 35MB to 50MB per minute.

Networking Integration
High-end Enterprise NAS usually adopt two networking interfaces to separate the loading of network and data storage that can balance the efficiency in a large-scale network infrastructure.

iSCSI application and integration
iSCSI devices offer major benefits to rugged and industrial environments by ensuring the reliable storage of mission critical information while reducing costs and maximizing resources. A filing system is usually File Level storage to store the data but using storing level functions. Some servers like a SQL serve are not able to load the file in shared directories. When iSCSI is a Blocking Level Storage which makes sure that all data can be accessed with reasonable performance by directing the storage block addresses where the data is going to be placed so that iSCSI can be operated in local server and the SQL server can easily access the data/files. It is a method of high efficiency and cost effectiveness.

Future Development

Even the NAS has been a mature product in business, the demand of larger and larger storage capacity and access to the digital data through wire or wireless for consumers, NAS is also ready for end user market. The growth will be explosive and it is expected that NAS will become the next generation media gateway.

According to most of enterprises with which the writer has spoken, the average spending of data storage equipment is accounted for 5% of IT budget. If adoption of SAN, the expending might exceed 50% of IT budget. The cost difference between NAS and SAN is apparent. For cluster system, the cost of an iSCSI NAS is also more reasonable than a FC SAN that can save a great purchasing cost.

Flowing with this trend, big-size businesses might still use FC SAN in crucial data storage areas and NAS for sub area in the future. The combination can ensure the security and reliability and cost effectiveness as well. The small-medium businesses might not be able to afford SAN system cost hundred thousands but use high C/P NAS in their data storage systems. The high-end NAS also support iSCSI that can expand the application of NAS. The percentage of NAS for home users is not obvious at this stage but the writer believes the usage of NAS for home users will be more in the future. When Windows VISTA and Windows Server 2003 and above all support iSCSI as a complimentary tool, iSCSI will become the main stream. In terms of manufacturers, Taiwan-based D-Link, Synology, Promise, QNAP Systems all introduced powerful and reliable NAS equipments. It is worthy believing, the popularity of NAS in enterprises is visible. Besides coping with the iSCSI technology, the iSCSI Initiator was embedded in Windows Vista which was introduced in 2007 by Microsoft that saves extra time to download. The Windows server 2008 also did the same thing. Jade Quantum Technologies, a Taiwan-based iSCSI manufacturer, shares a good chance in growing in this area.

In an enterprise setting, many companies use routers and firewalls a half of which are installed in a high-efficiency personal computer or server. The loading of a PC or server-structure usually performs better than other hardware devices that is a common solution for handling big amount of data. However, no matter how big CPU and memory is how high-end the hardware is, the network card is usually neglecting and the computer gets easily down when adopting cheap networking gateway. The main reason for the computer to be down is not because of the mainframe but because the over loading for the network card. Besides, the quantity of network card matters. A router might include three segments; therefore three PCI cards should be plugged that will cause a lack of efficiency and occupied slots.

Differences among network cards

Some told the writer that the network card enables a computer to connect networks so will it be different to use a US$5 network card or a US$50 one? A higher priced network card can undertake higher loading and the best advantage of it is less CPU utilization and higher stability to prevent the crash. The below is the CUP utilization comparison of two network cards.

Resources: NIC LAB, http://www.geocities.co.jp/SiliconValley/6445/

To select a good network card can sustain the performance of backbone gateway and network. Moreover, the high efficiency will keep the network flow to elevate working competence. Therefore, to spend US$45 more is worthy.

The application of Multi-Port card

Even a single high performance network card can work steadily, the slots might be insufficient in a PC. The multi-port card can be a solution to save this problem. A multi-port card is a good fit for a backbone network or gateway. Each single chip multi-port serial card can be detected in the operating system with individual reconfiguration and reboot of the individual partitions. Not only a better resource utilization but also an effective use of slot and easy to install, many enterprises are framed the backbone network in this way. It might take hundred or thousand US dollars to achieve the same performance but a multi-port serial card adding with the cost of PC might not more than US$300.

Resource: http://www.computex.biz/portwell/default.aspx?PageType=ProductDetail&pdt_id=15121&pcode=JW200

The best choice in the economic crisis

When the global economic recession continues, enterprises’ IT budgets are cut down in half. With the limited budget, it is the best timing for IT department shows its value to elevate the IT performance. A full-function unified threat management (UTM) will cost US$4,000 but it will only take US$700 to install the firewall on Linux Base. The price difference can be up to six times but the performance result is the same. For some examples that the write knew enable a 4-port PC to become an UTM plus the solution for three segment routers. When the multiport serial card has been a mature product, it will definitely create the new IT values for enterprises in this cost sensitive time. Taiwan-based multi-port serial card manufacturers include Portwell Inc., Ruby Technology Corp., Lycom Technology Inc., Ovislink Corp. and Wynn Tech Co., Ltd. The writer believes these manufacturers can offset the economic crisis to grow just like on-line game industry.

With more and more introductions of Netbooks and high-end laptops such as Lenovo X13, the evaluated advantages of solid-state drive (SSD) are getting well-known that also drives more users to chose SSD. However, SSD is not only used in notebook and netbook but also a good choice for system partition of desktop or server, the editor noted.

The advantages of SDD (solid-state drive) comparing with HDD (hard disk drive)

1. non-volatility and less power consumption
2. along with the lower power consumption, much lesser heat dissipation for systems
3. to withstand extreme shock and vibration with data integrity and without any danger of data loss
4. randomly accessed data without relying on a read/write interface head, short response time

In reality, the room occupancy, power consumption, software fee and upgrade/maintenance and reliability are major indexes on purchasing decision of server hardware. The above mentioned point 1, 2 and 4 are major elements to purchase a server in terms of cost effectiveness. The power consumption can reflect on lower electricity bill and the anti-shock feature at the 3rd point fits for the development of embedded market. The advantages of SSD are apparent.

On the contrary, couple disadvantages of SSD are as followed:

1. high cost per megabyte of storage
2. low popularity
3. lifespan has been questioned and criticized. (More explanations in the following paragraph

When I first time saw the dissembled SSD, the impression of SSD was like a RAM-DISK with a great amount of Flash Rom on the circuit board or like a big piece of flash drive. This is also the reason why SSD shares greater physical resilience to physical vibration, shock and extreme temperature fluctuations and the response time is almost 0 second.

Source: http://www.geckoandfly.com/tag/32gb-ssd

The differences among SLC, MLC and SLC+MLC

hen the traditional electromechanical hard drive is categorized by spinning speed, capacity and memory buffer, SSD is specified by the use of chip. The advance spec is SLC standing for single-level cell with highest speed and longest lifespan but higher cost per megabyte of storage due to difficult manufacturing process. Comparatively, the lower priced MLC for multi-level cell was slower than SLC in terms of speed and lifespan. As the gap between SLC and MLC is still great, several companies start to introduce the SDD employed with SLC+MLC for consumer to chose. The editor has arranged a comparison chart among SLC, MLC and SLC+MLC.

The write/erase endurance does not mean the SSD lifespan will be over when the endurance hits 10,000. The endurance does not count one time when each time to write/erase a file, either. It is counted as a page which is about 2KB. Each time to write data in SSD, the previous data need to be erased so that the speed of SSD is slower than other drives. As erasing the data, it is counted by 128KB which equals 64 pages. The write/erase endurance means the frequency of writing and erasing in each block. SSD wear-leveling technology enables the data to be written evenly in each block.

SSD provides faster system speed.

SSD’s pricing is still high but the high efficiency is what users want. Here is the mix of SSD and traditional electromechanical hard drive for the system to be installed. The advantage of SSD’s high speed and the big capacity and durability of traditional hard drive ensures the processing speed and data storage reliability that can decrease the waiting time and increase the working efficiency. The following video clip is shown in dual screens representing SSD and HDD respectively. The testing comparison is made by a Japanese engineer. The testing will help readers to understand how SSD improves working efficiency.

Afterword

After viewing the video, it is obvious to tell the high speed of SSD applied in running system. Taiwanese makers including A-Data, Apacer, Transcend and PQI have introduced different SSDs on the market. It is noteworthy that IBM and HP have SSDs available for blade servers in order to elevate processing speed and efficiency when HDS employs SSD in enterprise storage market. EMC Corp. announced an upgrade to its entire line of Celerra network-attached storage systems, saying it offers solid-state disks and file-level data de-duplication on the arrays and its gateway device as China-based Dexin Corp. also introduces SSD servers for ready shipment. It is estimated that SSD’s maturity and popularity and lower price can become mainstream of storage device of servers and desktops.

The editor believes the factors make SLC SSD a good fit in server/embedded systems, while MLC SSD makes it possible to create affordable mobile/netbook devices with large amounts of data storage when the cost of SLC SSD is three times more than MLC SSD’s. Therefore, to lower the cost of SSD will majorly depend on NAND Flash’s pricing. Also, if Taiwanese makers want to decrease the entire cost, the cost of SSD other components should be concerned. Here is the analysis of SSD components versus Taiwanese suppliers.

• Controller can be divided into Flash controller, Sata controller, RAID.Suppliers include Phison Electronics, Silicon Motion, JMicron, InCOMM Technologies and Skymedi Corp. The supplier of EeePC controller should be Silicon Motion. Besides, to integrate Flash controller and SATA bridge in one chip is the future trend. If Raid controller can be integrated in the same chip, SSD will be more available for server related products.
• NAND Flash: it is the primary use in SSD. The current suppliers are overseas makers. However, the 12-inch DRAM plants are gathered in Taiwan that will be the strength of Taiwanese makers. Makers include Powerchip Semiconductor, Transcend and A-Data.
• PCB, Passive component, Case: tons of Taiwanese manufactures are beyond description. For these are not the key components of SSD, the article is not going to specify in further details.