KAUST Repository Cloud Computing in KAUST Library: Beyond Remote Hosting Item type Authors Journal Conference Paper Yu, Yi Global Information and Management Symposium (GIAMS) Downloaded 29-Aug-2015 03:51:34 Item License Link to item http://creativecommons.org/licenses/by-nc-nd/4.0/ http://hdl.handle.net/10754/313410
[Cloud Computing in KAUST Library: Beyond Remote Hosting] 1 Cloud Computing in KAUST Library: Beyond Remote Hosting Yi Yu King Abdullah University of Science and Technology, Saudi Arabia Author Note Yi Yu, Library of King Abdullah University of Science and Technology, Saudi Arabia Yi Yu is a Systems Librarian of the library of King Abdullah University of Science and Technology in Thuwal, Saudi Arabia. This paper is presented for the international conference: Global Information and Management Symposium (GIAMS) on December 5-7th, 2013 in Taipei. Contact: yi.yu@kaust.edu.sa
Cloud Computing in KAUST Library: Beyond Remote Hosting Abstract Enterprise computing is the key strategic approach for KAUST to build its modern IT landscape. In such a strategic direction and technical environment, the library tries to establish library technology by catching new trends which help to make the library more efficient and sufficient. This paper focuses on the cloud computing development in the KAUST library, by using real world scenarios and first-hand experiences to describe what cloud computing means for KASUT library. It addresses the difficulties that were met by the library during the implementation process, how cloud computing affects the functional performance and work procedure of the library, how it impacts the style and modal of the library s technical service and systems administration, how it changes the relationships and cooperation among the involved players (the library, campus IT and vendors), and what the benefits and disadvantages are. The story of cloud computing at KAUST will share the knowledge and lessons that the KAUST library learnt during its development, and will also point out the future direction of cloud computing at KAUST.
[Cloud Computing in KAUST Library: Beyond Remote Hosting] 3 Introduction King Abdullah University of Science and Technology (KAUST) is a recent science university which officially opened in Saudi Arabia in September of 2009. Named after the nation s leader who had a vision for a new House of Wisdom that would become one of the world's great institutions of research, the KAUST university removed its scaffolding and cranes to reveal an ultra-modern campus on the shore of the Red Sea on the site of an ancient fishing village called Thuwal beside the thriving metropolitan city Jeddah. Just like its parent university, KAUST Library is a brand new library which was built from scratch. During 4 years of development, KAUST library successfully grew into a medium sized academic and research library which includes 25 staff members, over 10,000 print collections and about 50 subscriptions of electronic resources covering the major science databases and e-journals. Besides its presentation of a new generation library by its spacious and artistic modern architecture, the library is also equipped with the most advanced facilities and library systems. As enterprise computing is the key strategic approach for KAUST, it naturally becomes a direction towards which to build the KAUST library s modern IT landscape. In such a strategic and technical environment, KAUST library tries to establish library systems by utilizing new trends and maximizing the benefits of new technology. Going towards cloud computing is one of these efforts and achievements. What is cloud computing and why go to cloud? The term cloud computing is more like a metaphor instead of an accurate and literal definition. As the word cloud" conventionally refers to "the Internet", the phrase cloud computing presents "a type of Internet-based computing" where different services -- such as servers, storage and applications -- are delivered to an organization's computers and devices via the Internet. i One of the reason that cloud computing has become so popular can be attributed to its ability to supply host services, which refers to application services that run client server software from a remote location. ii The application of cloud computing at the KAUST library is focused on this
Cloud Computing in KAUST Library: Beyond Remote Hosting aspect. However, the experiences and lessons that we have achieved does not only apply to the topic of remote hosting service. From the very beginning, as one of the major server applications in the campus, KAUST library system was hosted locally in KAUST IT Data Center. When cloud computing became a new approach for campus IT infrastructure, we were thinking of the possibility and feasibility of using cloud computing for library systems. After discussions, we decided that since internet connections are no longer a barrier, there are many benefits for the library to apply cloud computing in our surroundings. First of all, by switching to remote servers hosted by our vendor, we are able to have more dedicated and professional technical support for the library. Secondly, full package services such as upgrading, database maintenance and backup will allow the library s technological staff to do other strategic activities rather than routine operational and maintenance tasks. The benefits also include reducing mistakes that arises from misunderstandings between venders and campus IT, increased speed and efficiency for troubleshooting and problem solving, among many other advantages. In addition, due to special security concerns in this unique university, putting all of the ILS application in the cloud will be beneficial for the security of the entire campus network. How to implement? The first step of implementation was the migration of all servers to the vendor hosted location. As there are many tasks to be done -- organizing the migration schedule, changing DNS entries, opening ports for the remote host, updating IP addresses for electronic resources etc. -- the process required the involvement and corporation of many units among the library as well as the campus IT and the host vendor. In order to have a successfully migration, many people from different departments needed to work together. Even though we employed extensive communication and coordination in preparation, as well as tried to predict any potential problems and to anticipate solutions, there were still unexpected issues and situations that occurred during the migration process. For example, the window of time for the server shutdown was selected based on the KAUST IT CIO s suggestion and was agreed to by the three major players involved in the process the vendor, the campus IT and the library. However, it incurred many complaints
[Cloud Computing in KAUST Library: Beyond Remote Hosting] 5 from the KAUST reference department because they had scheduled a tour for new employees during this time, and had arranged for a demo of the library s online catalog during that day. Another unforeseen incident during the implementation process was the special security procedures employed by the KAUST campus network. Many ports in the KAUST campus network are closed unless there had been a request to open it through certain procedures, and this is a situation that the host vendor has never experienced before, which resulted in the library s online catalog to be unable to be accessed even though there were no connection problems between the server and the client applications. This caused a big panic in the library. Once we realized what the issue was, the system librarian coordinated between the KAUST IT and the host vendor to resolve this problem. Since updating IP addresses of electronic resources was not included in the migration to do list from the vendor, we did not anticipate informing the e- publishers to change the IP address accordingly. After the implementation, none of the e- resources could be accessed. The result was that the migration was a total failure, and we had to roll-back to the old server. When everything seemed to be proceeding well for the second attempt of server migration, we ran into a very strange situation with the self-check machines. Normally, we can update the new server IP in the self-check machine through the admin interface. However, the admin interface could not be displayed. After wrestling with this issue for a couple days, we found that the issue was caused by the installation of a new anti-virus software which blocked the display. This is a very unexpected and strange coincidence since we initially thought that it was due to the server migration that caused the issues with the self-check machine, but in the end realized that it is due to a completely unrelated situation. Even when all the connections in the public interface looked normal, there were still many emergent complaints pouring into the system librarian: many staff members of the library were not able to establish any client-server connections. In the end, we found out that this was actually caused by a space that was accidentally left in the new IP address when they were typing it into the configuration file. Such panic and chaos appeared one after another during the initial days after the implementation process. Failing miserably in a very easy task is not a cliché. In this situation, taking great care in the preparation, sincere corporation and coordination, and detailed communication are the key factors for making the implementation process smooth and easy. Certainly, having solid technical knowledge is also a great help in getting things done.
Cloud Computing in KAUST Library: Beyond Remote Hosting What are the problems in the cloud? After our integrated library systems was moved to the remote datacenter hosted by the system vendor, the library staff members found that their client application would automatically disconnect if it was left idling for even a few minutes. This disconnection issue had never occurred when the servers were hosted locally by the KAUST IT. The staff members routinely use many different modules of the application to perform tasks such as cataloging, circulation and acquisition, and this issue was very annoying for the staff since they would need to constantly restart the module even if it was only inactive for a couple of minutes. Figure 1 - disconnection message from client application of library system In order to debug this issue, the KAUST IT network team installed a network packet analyzer called Wireshark on several workstations. The analyzer was used to closely monitor and record the pattern of disconnections, to help us understand the transaction behavior, and also to capture
[Cloud Computing in KAUST Library: Beyond Remote Hosting] 7 the packet data for identification of the cause on the packet and protocol level. After the issue was investigated based on the captured data, the inter-network situation seemed to become clearer: The disconnection occurs intermittently and in a random pattern; The Client-Server communication uses a keep-alive mechanism, on ports other than the ports used by the SCCP protocol; Protocol inspection for the SCCP protocol was required by the application vendor, but was turned off on the firewall by KAUST IT; Packet captures on both the server side (Internet) and the client side (Inside) had shown TCP RST received on both sides. This information acquired shows that a device in between is sending RST frames in both directions to terminate the session after a certain timeout period. IT Security increased the TCP session idle period to 8 hours (60 minutes is the default). However, the disconnection is happening frequently and the sessions would end well before 8 hours, so this cannot be the cause iii of the problem. How to resolve? In order to figure out which part of the chain is causing the disconnection, KAUST IT examined potential nodes by identifying internal and external traffic paths.
Cloud Computing in KAUST Library: Beyond Remote Hosting Figure 2 - traffic path, inside is marked green while outside is marked red. The nodes to be tested in the outside path are: International Gateway (IGW) KAUST MPLS Coreiv The nodes to be tested in the inside path are: Firewall Anagranv F5 Load Balancervi
[Cloud Computing in KAUST Library: Beyond Remote Hosting] 9 An exclusive method was used to investigate this issue. The results showed that all the test machines placed right before IGW, the Virtual routing and forwarding (VRF) vii, and the firewall were not experiencing the disconnections, but the reference machine on the Data VLAN was experiencing it. Thus it was reasonable to conclude that the disconnection was being caused by the F5 load balancer. After adjusting the settings in F5, the disconnection problem has been resolved. Lessons and next step There are several points that are worth noting about our implementation of cloud computing. First, and the most crucial step, is to make a correct decision about whether or not to switch to cloud computing. As there are many debates about the pros and cons of cloud computing, it is necessary to investigate and analyze the whether there is a need in reality before making the decision. Considering that there are many units from different departments and organization who will be involved in the process, it is also extremely important to obtain full understanding and support from all related departments and personnel. During the process of implementation, each step and every detail should be considered thoroughly. Any neglect or carelessness can lead to big disasters just like what we experienced, resulting in a complete rollback to the old server. Some of the problems were not avoidable as it was a new experience for us and we could not anticipate all the potential issues. As everything is in the cloud now, network and security issues have become a prominent topic, as things may have become more complicated and invisible. This does not only bring new challenges to the KAUST IT, but also raised new questions to the server application -- such as how the system has adapted to the new network environment. For example, during the investigation period for the disconnection problem, the server/host vendor and KAUST IT engaged in long arguments about who should take responsibility for the disconnection. Neither party thought that the problem should be their responsibility. Even though the final result indicated that the problem was caused by the configuration of F5 Load Balancer, the KAUST IT still believes that the issue lies within the applications for not being updated to adapt to a remote hosting environment, because this problem did not appear with other remotely hosted applications running on the same F5 configurations. The process of implementing cloud computing in the KAUST library has brought
Cloud Computing in KAUST Library: Beyond Remote Hosting many new challenges to be tackled because this is a new experience for everyone involved the KAUST IT, the system librarian, the application vendor and every component had to become adapted to the new network environment. As cloud computing is a new practice for the hosting/application vendor as well as the KAUST library, many things will have to change according to the new situation. There are still a lot of unclear issues and gray areas. We need to review all of the documents and protocols including contracts, agreements, emergency procedures, etc. We will need to discuss with the suppliers, vendors and KAUST IT regarding how to revise and complete the documents based on the new technical environment. For the systems librarian, cloud computing brings with it a new and unique set of tasks and responsibilities. Instead of only focusing on the functionalities and performance of the library system, the systems librarian will also need to take care of the whole chain of the system which spans across the border of both the internal and external networks. Thus, the systems librarian needs to have a broader vision regarding the organization and the technology, a better understand about the breadth of different technologies and platforms, to have the capability to think outside the box, and to also learn the language of business. Knowledge about networks and network security is also an important and prominent area for the system librarian in the cloud computing era. Conclusion The KAUST Library decided to move its integrated library systems to a remote host in following with the cloud computing approach adopted by KAUST IT. The process of server migration to the remote host was a major undertaking. Careful preparation, full cooperation, and smooth communication are key attributes for the success of migration. During the initial stages of working with cloud computing after server migration, we encountered various difficulties and problems including client-server disconnections. We used the exclusive method to investigate session disconnection issues and found that the setting of the F5 Load Balancer was the cause. We will continue to work with KAUST IT and the application vendor and the host supplier to improve the service and management of the KAUST library systems. The library technical staff will continue to acquire new knowledge both in technology and business in order to catch new trends of computing service and to continuously improve the library.
[Cloud Computing in KAUST Library: Beyond Remote Hosting] 1 1 Reference i Webopedia. N.p., n.d. Web. 13 Nov. 2013. <http://www.webopedia.com/term/c/cloud_computing.html>. ii "Cloud computing." Wikipedia. Wikimedia Foundation, 18 Nov. 2013. Web. 18 Nov. 2013. <http://en.wikipedia.org/wiki/cloud_computing>. iii Report of Millennium Problem, KAUST IT, Aug. 2013. iv "Multiprotocol Label Switching." Wikipedia. Wikimedia Foundation, 18 Nov. 2013. Web. 19 Nov. 2013. <http://en.wikipedia.org/wiki/ Multiprotocol_Label_Switching v "Anagran." World's Largest Professional Network. N.p., n.d. Web. 19 Nov. 2013. <http://www.linkedin.com/company/anagran>. vi "Load Balancer." Home. N.p., n.d. Web. 19 Nov. 2013. <http://www.f5.com/glossary/load-balancer/>. vii "Virtual Routing and Forwarding." Wikipedia. Wikimedia Foundation, 17 Nov. 2013. Web. 19 Nov. 2013. < http://en.wikipedia.org/wiki/virtual_routing_and_forwarding>