In network design, availability of the store information is dependent on the storage system in which it is stored and the online connectivity of those systems. In attached disk systems, the two factors show sufficient reliability and users need only to handle the issue of availability in an ad hoc manner when making regular backups. With RAID systems, the probability of disk failures is reduced to a certain degree. However, connectivity challenges are not affected by it. This is so because hardly do catastrophic network failures happen between a disk and the processor attached to it.
Storage Area Networks (SANs) have taken the option of working with directly attached disk model, taking the assumption that connectivity will never fail at least, not in catastrophic levels. This brings to the question of the availability of SANs within the wide area network (WAN).
It is possible that in Wide Area Networks, storage mediums can be unavailable intermittently or when they are available, the quality of service could be inadequate. This is usually caused by a number of factors in the network including congestion in the network, problems in routing, changes in topology and sabotage in the form of malicious interference.
Resolution of these kinds of network problems can take between seconds to several days. Therefore, it is necessary to develop end to end strategies that will ensure continued availability. In the case of Alpha Omega Technical Industries given its rapid growth, a wide range of strategies will be put in place to ensure availability. This will range from simple automated retries to redundant data accesses distributed across the entire network. This will be augmented with a kind of RAID error correction system to minimize data redundancy. These two measures will serve to ensure high latency in archival backups. The implementation of the algorithms to enforce the above measures will be done at end points, to ensure delivery to the end point and to get to the required level of sensitivity requirements of the operating systems at the end points, applications and users. The Venti design (Venti, 2002) where storage block are indexed by 160 bit hashes of their data is the right approach to ensure the unification of the various end to end strategies, hence, ensuring maximum system availability.
System Accuracy and Correctness
The assumptions as concerns the accuracy and correctness of storage systems have evolved over the years. At some point in the evolution, the tape was viewed as unreliable as a result of various environmental factors while at the same time, the optical disk was considered reliable. As a result, verification of data stored in tapes was done using checksums while disk data was never verified. This was based on the assumption that disk storage fails by the corruption of entire sectors which was easy to detect by the hardware controller. However, with the emergence of mass produced disks a wide variety of complex failure modes such as the undetected delivery of incorrect data has been common (Quantum, personal communication 2001).
In Storage Area Networks, it is assumed that RAID storage systems have adequate internal checking capabilities to ensure accuracy and that the networks utilizes highly reliable network protocols. Given that Storage Are Networks are implemented in extremely controlled environments, the components within the network are assumed to be reliable. This can or cannot be a reasonable assumption, however, for Alpha Omega Technical Industries; there is a need to ensure an end to end approach to the WAN requiring data accuracy to be checked by the end systems. This means that the ultimate writers and readers of the data will always be checking for data accuracy. This will not only ensure protection against errors in the composition of a reliable disk and unreliable network components but also against malicious components that could be introduced without being detected from the wide area network (WAN). In this situation too, the implementation of the Venti Design can also prove to be beneficial given that handle to data is a checksum of the data.
The storage systems that are directly attached and localized storage area networks solutions are considered to have a high level of physical security that can only be breached by highly complicated intrusive means. However, once the storage area networks are connected to the office networks and especially when tunneled through the WAN, the security definitely becomes an issue. When it comes to Internet Protocol (IP) networking, one cannot assume that intermediate nodes are trustworthy. Therefore, for Alpha Omega Technical Industries, there is a need to implement an end to end application of cryptographic systems.
Nonetheless, the security aspect of Alpha Omega Technical Industries cannot sufficiently manage Denial of Service with just end to end techniques alone. In this regard, the security mechanisms implemented in the network should have control over the rights to allocate storage. Even then, it must be appreciated that any kind of security measures are just best efforts and the end points must be fully prepared to be affected by denial of service attacks, which might render data allocations within the WAN impossible or make it possible to have stored data overwritten as a result of security breaches at intermediate nodes. The techniques for detecting or preventing denial of service attacks are the only bits in the network design that cannot be handled in an end to end manner. However, analogous storage can be implemented to handle Denial of service in IP networks.
Conclusions and Recommendations
The end to end network security argument was developed in the context of IP network architecture, which encapsulates the end to end delivery of datagrams from sender to receiver. This network system conceptualization is highly scalable, and storage sharing can be designed in line with end to end principles. The resultant architecture creates a model for the extension of storage networks into the Wide Area Networks. This is an exciting niche in the development of storage technology.
It is advisable for Alpha Omega Technical Industries that the deployment of end to end networking solutions is the most effective way as they grow. Therefore, they need to scale their data storage systems. In any case, wide spread deployment of storage internetworks will hugely improve the capacity and the flexibility of the network. In using this model, the network architecture evolves into radically decentralized computer architecture.
Quinlan, S. and Dorward, S., Venti (2002) a new approach to archival storage. in USENIX File and Storage Technologies. Monterey, CA, 2002
Reed, D.P., Saltzer, J.H. and Clark, D.D. Comment on Active Networking and End-to-End Arguments. IEEE Network, 12 (3). 69-71.May/June
Plank, J.S., Bassi, A., Beck, M., Moore, T., Swany, M.and Wolski, R. (2003) Managing Data Storage in the Network. IEEE Internet Computing, 5 (5). 50-58.