Feedback

A Tunable Checkpointing Algorithm for Distributed Mobile Applications

Documento informativo
IJCSI International Journal of Computer Science Issues, Vol. 8, Issue 6, No 3, November 2011 ISSN (Online): 1694-0814 www.IJCSI.org 1 A Tunable Checkpointing Algorithm for the Distributed Mobile Environment Sungchae Lim Dept. of Computer Science, Dongduk Women’s University Seoul, 136-714, South Korea Abstract The aim of a distributed checkpointing algorithm is to efficiently restore the execution state of distributed applications in face of hardware or software failures. Originally, such algorithms were devised for fixed networking systems, of which computing components communicate with each other via wired networks. Therefore, those algorithms usually suffer from heavy networking costs coming from frequent data transits over wireless networks, if they are used in the wireless computing environment. In this paper, to reduce usage of wireless communications, our checkpointing algorithm allows the distributed mobile application to tune the level of its checkpointing strictness. The strictness is defined by the maximum rollback distance (MRD) that says how many recent local checkpoints can be rolled back in the worst case. Since our algorithm have more flexibility in checkpointing schedule due to the use of MRD, it is possible to reduce the number of enforced local checkpointing. In particular, the amount of data transited on wirelesses networks becomes much smaller than in earlier methods; thus, our algorithm can provide less communication cost and shortened blocking time. Keywords: Mobile networks, distributed application, rollback, recovery, distributed checkpointing. 1. Introduction During the past decades, there have been dramatic advances in mobile networks and mobile devices. In particular, the fast spreading usage of smart phones is likely to yield demands for sophisticated distributed applications across multiple mobile devices [1, 2]. During the run-time of such a distributed application, its cooperating application processes (APs) work in parallel and data are usually transited between APs to share application contexts. In this situation, failure on a single AP or hardware device could cause a serious problem in the whole distributed application and thus it may roll back the application’s processing state to the initial one in the worst case. To prevent a whole cancelation of the processing result, checkpoint records are created to log intermediate execution results. The recovery procedure after abrupt failure builds a consistent state of an application from the checkpoint data, and resumes the interrupted application from that state. This can reduce undesirable loss of application process To make the distributed application robust and recoverable against failure, many works are done for the computing environments where the distributed application seems to be executed in the wired fixed networks [4, 6, 7, 10, 11, 13]. When checkpointing algorithms of those earlier works are applied to distributed applications running on wireless networked, they suffer from high cost for sending checkpoint data via warless connections. Since data transit over wireless networks is more costly and unstable, compared with that over wired networks, many researches focus on reduction of wireless data transit in the case of the checkpointing scheme for wireless computing environment [5, 8, 9,10, 12]. In the paper, we also propose a distributed checkpointing scheme suitable for distributed applications running on the mobile computing environment. We here introduce two key ideas of the maximum rollback distance (MRD) and the logging agent running on the MSS (Mobile Support Station). The logging agent is a software agent running on MSS, which is responsible for making local checkpoints at the request of its associated AP’s requests and maintaining at least one consistent global checkpoint. To save the cost for maintaining such a global checkpoint, the agent can do some logging activities without any requests from associated AP. For this, the logging agent securitizes messages arriving on its MSS and communicates with other logging agents for synchronization of checkpointing. IJCSI International Journal of Computer Science Issues, Vol. 8, Issue 6, No 3, November 2011 ISSN (Online): 1694-0814 www.IJCSI.org If the rule of checkpointing synchronization is too strict, enforced local checkpoints are frequently created. Sine enforced local checkpoint request costly data transit in wireless network lines; it is needed to make the synchronization rule more flexible. In this notion, we introduce the MDR for each distributed application. On the other hand, the MDR is a run-time parameter for a distributed application, saying how many local checkpoint of a given AP can be rolled back in the worst case. With a MDR properly set to a value, a significant flexibility is available at the time local checkpoints are synchronized in order to create a new global checkpoint. Due to the tunable level of checkpointing synchronization using MDR, the logging agents participating in a distributed application can reduce the number of enforced local checkpointing and costly message transit over wireless networks. The rest of this paper is organized as follows. In Section 2, we describe some backgrounds regarding the meaning of global consistency of distributed checkpoints, the assumed mobile network, and the previous works. Then, we propose a new efficient distributed checkpointing scheme in Section 3, and discuss the performance characteristics of our scheme in Section 4. Lastly, we conclude this paper in Section 5. 2. Backgrounds 2.1 Global Consistent State The GCS (Global Consistent State) of distributed applications was formally defined by Lamport [17]. According to that definition, processing of a distributed application can be modeled by three types of events such as the message sending event, the message receiving event, and the computation event. Each AP participating in a distributed event can do the computation event to proceed with its processing state and communication with other participant APs through message sending/receiving events. In this event model, a set of events meeting the GCS can be captured using the relation “happen-before” drawn on events. In [17], the “happen-before” relation (HBR) is as follows. [Definition of HBR] If it is the case that e1 “happen-before” e2, then either of the following conditions should be true. i) Both e1 and e2 occur in the same AP and e1 precedes e2 in timing sequence. ii) There are a message m and two APs of p1 and p2 such that p1 sends (event e1) message m to p2 and p2 receives (event e2) it. 2 Fig. 1. An example of GCS: a failure arises at p3. Owing to the transitive property of the HBRs, we can give a partial order to the events of a distribute application, even though there is no common clock shared by its participant AP’s. The GCS of a distributed application is defined based on the HBR above. Let us take a snapshot of execution state of a distributed application at a particular time, and let S be that snapshot, which is a set of the events having arisen in the application. Let G be a subset of S. In this case, G is said to be in a GCS if the following condition is satisfied; for every event e’ in G, if there is e in S such that e “happen-before” e’, then e should be also an event in G. In other words, for every event of G, its causal events should be found in G. Since all the causal events are contained in G, it may be possible to obtain the same execution results of G, if we redo the events of G from it begging time. Based on this idea, we can recover any intermediate execution state of any failed application if its any GCS execution snapshot is available. To have execution snapshots, checkpointing schemes are commonly used for saving local execution state of individual AP’s. Fig.1 shows an example where distributed checkpoint is performed by three AP’s, p1, p2, and p3. In the figure, the blacked rectangle of C i,k represents the k-th local checkpoint made by AP p i . The local checkpoint of C i,k is made to save the computational computation state of p i and the message sent to other AP’s after the creation time of C i,k-1 . Suppose that an application failure arise at p3. as in Fig. 1. At this moment, the set of local checkpoints preserving the GCS are that inside the GCS line of the figure. That is, the latest CGS state is composed of C 1,3 , C 2,2 , and C 3,3 . As the message sending event of m8 is not saved in any local checkpoint, its message receiving event cannot be include a GCS. Therefore, C 2,2 is rolled back, and C 1,3 is also rolled back because C 1,3 contains the message-receiving event of m 7 saved in C 2,2 . As a result, the local IJCSI International Journal of Computer Science Issues, Vol. 8, Issue 6, No 3, November 2011 ISSN (Online): 1694-0814 www.IJCSI.org 3 checkpoints on the GCS line will be used to recover the failed application. In the example of Fig. 1, the latest local checkpoints of a GCS are the same as C 1,3 , C 2,2 , and C 3,3 is used as a latest consistent global checkpoint. This choice of such a consistent global checkpoint is done by a recover algorithm initiated in the presence of failure. During the recovery phase, such a latest consistent global checkpoint is found and then the disrupted distributed application is restarted. In the case of Fig. 2, the associated AP’s will restore their computational state using the data saved in the local checkpoints of C 1,3 , C 2,2 , and C 3,3 , respectively, and p1 will send the lost message m 6 to p2 again. Since checkpoint records bookkeep the serial numbers of messages transferred among AP’s, this message resending is possible. 2.2 Assumed Mobile Network In general, the mobile network is comprised of mobile hosts (MH’s), mobile support stations (MSS’s), and the fixed networks interconnecting the MSSs [1, 2, 8]. The network architecture is shown in Fig. 2(a), where there are two wireless cells and MH’s can make wireless network connections within its wireless cell. Since the MH can hop among wireless cells, the MSS’s have to update the list of MH’s under control for seamless hand-offs. Each AP can be identified by unique process id within its hosting MH. Of cause, the MH is also uniquely identified in the global network environment. Consider a situation where a MH x in MSS A sends a network message m to an MH y in MSS B. The message m from x is queued into an outbound queue of MSS A and then it is delivered to the counterpart MSS B via the fixed network. Consecutively, message m is entered into the inbound queue of MSS B for the delivery towards y. On the top of the traditional architecture of Fig. 2(a), we assume that an agent program executes on each MSS for doing checkpoint-related activities. That is, it is assumed that the agent program makes accesses the two message queues of the MSS, in which outbound or inbound network messages are temporarily stored waiting for their delivery to target AP. Fig. 2(b) depicts the assumed architecture with logging agents. In the example of Fig. 2(b), an MH in cell A sends a network message m to other MH staying in cell B. In this case, the logging agent in MSS A dequeues message m and then appends some checkpoint-related data to m before it sends m to the logging agent of MSS B. Correspondingly, the logging agent of MSS B deletes the Fig. 2. Architecture of the assumed mobile network. appended data from m before it inputs m into the inbound queue towards the destination MH. During this message transit time, the logging agent’s can make checkpoint records in the disk storage installed in the MSS’s. Using the logging agent, we can reduce the checkpoint cost and improve the flexibility of the consistent global checkpoints. The more details about the logging agent are described in Section 3. 2.3 Earlier Works Checkpointing schemes for distributed applications can be roughly categorized into the synchronized schemes and the asynchronized schemes. In the synchronized schemes, when a AP requests a checkpoint, actions for making a consistent global checkpoint are performed such that the newly created global checkpoint includes the current execution state of the checkpoint-requesting AP. From this, the checkpoint-requesting AP can make its crucial results of execution robust to any failure. For such checkpointing, the checkpoint requester AP is blocked until all the causal events of the checkpoint-requested events are saved in the local checkpoint records of the participant AP’s. Owing to such creation of a global checkpoint, most of execution results can be restored in the present of failure. However, because creation of a global checkpoint needs a number of message deliveries and requires some enforced checkpointing of other participant AP’s, this scheme suffer from a high network cost and long delay time for checkpointing. Especially, such shortcomings become more serious in the case where the AP’s are ruing in mobile network environment [1, 2, 8, 12]. IJCSI International Journal of Computer Science Issues, Vol. 8, Issue 6, No 3, November 2011 ISSN (Online): 1694-0814 www.IJCSI.org Meanwhile, the asynchronized scheme does not enforce the creation of a consistent global checkpoint at every checkpoint request time. Instead, during the recovery phase a latest consistent global checkpoint is found from the casual dependency of APs’ events saved in disturbed local checkpoints. By examining the casual dependency among the previous local checkpoints, the recovery algorithm picks a most recent consistent global checkpoint for recovery. Since each AP can make its local checkpoint in asynchronized manner, this scheme is apt to have a problem of many cascaded rollbacks of local checkpoints, so-called domino effect [1, 8, 17]. In a worst case, the whole execution results of any distributed application can be cancelled because of the domino-effect. In addition, the asynchronized schemes have more restarting overheads, compared to the synchronized schemes. This is because the asynchronized scheme has to collect the whole information from scatted local checkpoints in order to find a consistent global checkpoint. From these reasons, the synchronized scheme is preferred in earlier time. However, when it comes to the mobile network environment, the synchronized scheme is more feasible because of less network connectivity and more consideration of instability of mobile devices of that application environment. Among the asynchronized schemes, in particular, the message-induced checkpointing scheme [8, 9] is regarded to be a good alternative solution in the mobile network environment. This is because the message-induced scheme can eliminate the possibility of domino effect sin a very simple manner. By forcing AP’s to make local checkpoints depending on the messagereceiving events, this scheme can set some boundaries on rollbacked local checkpoints 3. Proposed Method 3.1 Motivations Although the message-induced scheme is useful to avoid the domino effect, it has a severe problem in that the time of checkpoint creations is determined without active involvement of participant AP’s. This problem can be easily understood by viewing the used mechanism of the message-induced scheme. The scheme uses two different execution states of the AP, that is, SEND and RECEIVE states. The SEND and RECIVED states are set while the message-sending and message-receiving events are successively arising, respectively. A new checkpoint is compulsorily created at the time when a network message arrives at a particular AP with the execution state of SEND. Since the time of checkpoint creations in an AP is 4 inactively determined depending on the arrivals of message-receiving events, the created checkpoints would not reflect application’s semantics. Otherwise, if we want application’s semantics-aware checkpointing, that checkpointing time of checkpoint can be chosen by considering the critical points of processed application. In other words, checkpoints have to be made when some critical executions or expensive processes are done. Such semantics-aware checkpointing is possible only when the AP can actively request checkpointing. Note that the user can also issue checkpoint requests via its AP. The lack of semantics-awareness of the message-induced scheme may have poor performance. When this scheme is used for distributed mobile applications requiring a lot of message transit, a large number of non-meaningful checkpoints can be made. This can result in frequent checkpointing and creations of obsolete checkpoints. In addition, the message-induced scheme has no mechanism to actively create consistent global checkpoints. Therefore, in the case that an AP wants to make its critical execution results persistent, there is no way for that. Whether or not the execution results are saved into a consistent global checkpoint relies on the existence of an appropriate pair of message-receiving and message-sending events. To solve such problems of lack of semantics-awareness in checkpointing time and its defective mechanism for global checkpointing, we propose a new checkpoint scheme based on a combination of the logging agent and the R-distance 3.1 Data Format First, we describe the data format of the network message, which is represented by M below. In the followings, the message sender AP is denoted by and the total number of AP’s joining the distributed application by N, respectively. The fields of M are seven in all. Among them, the last three fields are not used by the AP. These fields exist for containing control data of checkpointing purpose and they are visible only to the logging agent. Meanwhile, the first four are for containing application data needed by the AP. M.type: Message type M.sender: Id of M.recever: Id of the counterpart AP M.data: Application data sent to the counterpart AP M.ap[1,2, ,N]: Ids of AP’s joining this application. enfocar la importancia de la producción mediática de los niños en su descubrimiento del mundo, sobre todo utilizando el periódico escolar y la imprenta. Asimismo las asociaciones de profesores trabajaron en esta línea e incluso la enseñanza católica se comprometió desde los años sesenta realizando trabajos originales en el marco de la corriente del Lenguaje Total. Páginas 43-48 45 Comunicar, 28, 2007 En el ámbito de los medios, también desde el principio del siglo XX hay ciertas corrientes de conexión. Pero es a lo largo de los años sesenta cuando se constituyeron asociaciones de periodistas apasionados por sus funciones de mediadores, que fomentaron la importancia ciudadana de los medios como algo cercano a los jóvenes, a los profesores y a las familias. Así se crearon la APIJ (Asociación de Prensa Información para la Juventud), la ARPEJ (Asociación Regional de Prensa y Enseñanza para la Juventud), el CIPE (Comité Interprofesional para la Prensa en la Escuela) o la APE (Asociación de Prensa y Enseñanza), todas ellas para la prensa escrita Estas asociaciones fueron precedidas por movimientos surgidos en mayo de 1968, como el CREPAC que, utilizando películas realizadas por periodistas conocidos, aclaraba temas que habían sido manipulados por una televisión demasiado próxima al poder político y realizaba encuentros con grupos de telespectadores. cipio del siglo XX, y nos han legado textos fundadores muy preciados, importantes trabajos de campo y muchos logros educativos y pedagógicos. La educación en medios ha tenido carácter de oficialidad de múltiples maneras, aunque nunca como una enseñanza global. Así la campaña «Operación Joven Telespectador Activo» (JTA), lanzada al final de los años setenta y financiada de manera interministerial para hacer reflexionar sobre las prácticas televisuales de los jóvenes, la creación del CLEMI (Centro de Educación y Medios de Comunicación) en el seno del Ministerio de Educación Nacional en 1983, la creación de la optativa «Cine-audiovisual» en los bachilleratos de humanidades de los institutos en 1984 (primer bachillerato en 1989) y múltiples referencias a la educación de la imagen, de la prensa, de Internet. La forma más visible y rápida de evaluar el lugar de la educación en medios es valorar el lugar que se le ha reservado en los libros de texto del sistema educa- 2. Construir la educación en los medios sin nombrarla El lugar que ocupa la edu- La denominación «educación en medios», que debería cación en los medios es muy ambiguo, aunque las cosas están cambiando recientemente. entenderse como un concepto integrador que reagrupase todos los medios presentes y futuros, es a menudo percibida En principio, en Francia, co- por los «tradicionalistas de la cultura» como una tendencia mo en muchos otros países, la educación en los medios no es hacia la masificación y la pérdida de la calidad. una disciplina escolar a tiempo completo, sino que se ha ido conformado progresivamente a través de experiencias y reflexiones teóricas que han tivo en Francia. Una inmersión sistemática nos permi- permitido implantar interesantes actividades de carác- te constatar que los textos oficiales acogen numerosos ter puntual. Se ha ganado poco a poco el reconoci- ejemplos, citas, sin delimitarla con precisión. miento de la institución educativa y la comunidad es- colar. Podemos decir que ha conquistado un «lugar», 3. ¿Por qué la escuela ha necesitado casi un siglo en el ámbito de la enseñanza transversal entre las dis- para oficilializar lo que cotidianamente se hacía en ciplinas existentes. ella? Sin embargo, la escuela no está sola en esta aspi- Primero, porque las prácticas de educación en me- ración, porque el trabajo en medios es valorado igual- dios han existido antes de ser nombradas así. Recor- mente por el Ministerio de Cultura (campañas de foto- demos que no fue hasta 1973 cuando aparece este grafía, la llamada «Operación Escuelas», presencia de término y que su definición se debe a los expertos del colegios e institutos en el cine ), así como el Minis- Consejo Internacional del Cine y de la Televisión, que terio de la Juventud y Deportes que ha emprendido en el seno de la UNESCO, definen de esta forma: numerosas iniciativas. «Por educación en medios conviene entender el estu- Así, esta presencia de la educación en los medios dio, la enseñanza, el aprendizaje de los medios moder- no ha sido oficial. ¡La educación de los medios no apa- nos de comunicación y de expresión que forman parte rece oficialmente como tal en los textos de la escuela de un dominio específico y autónomo de conocimien- francesa hasta 2006! tos en la teoría y la práctica pedagógicas, a diferencia Este hecho no nos puede dejar de sorprender ya de su utilización como auxiliar para la enseñanza y el que las experiencias se han multiplicado desde el prin- aprendizaje en otros dominios de conocimientos tales Páginas 43-48 46 Comunicar, 28, 2007 como los de matemáticas, ciencias y geografía». A pe- mente en todas las asignaturas. Incluso los nuevos cu- sar de que esta definición ha servido para otorgarle un rrículos de materias científicas en 2006 para los alum- reconocimiento real, los debates sobre lo que abarca y nos de 11 a 18 años hacen referencia a la necesidad no, no están totalmente extinguidos. de trabajar sobre la información científica y técnica y En segundo lugar, porque si bien a la escuela fran- el uso de las imágenes que nacen de ella. cesa le gusta la innovación, después duda mucho en Desde junio de 2006, aparece oficialmente el tér- reflejar y sancionar estas prácticas innovadoras en sus mino «educación en medios» al publicar el Ministerio textos oficiales. Nos encontramos con una tradición de Educación los nuevos contenidos mínimos y las sólidamente fundada sobre una transmisión de conoci- competencias que deben adquirir los jóvenes al salir mientos muy estructurados, organizados en disciplinas del sistema educativo. escolares que se dedican la mayor parte a transmitir Este documento pretende averiguar cuáles son los conocimientos teóricos. La pedagogía es a menudo se- conocimientos y las competencias indispensables que cundaria, aunque los profesores disfrutan de una ver- deben dominar para terminar con éxito su escolaridad, dadera libertad pedagógica en sus clases. El trabajo seguir su formación y construir su futuro personal y crítico sobre los medios que estaba aún en elaboración profesional. Siete competencias diferentes han sido te- necesitaba este empuje para hacerse oficial. nidas en cuenta y en cada una de ellas, el trabajo con Aunque el trabajo de educación en los medios no los medios es reconocido frecuentemente. Para citar esté reconocido como disciplina, no está ausente de un ejemplo, la competencia sobre el dominio de la len- gua francesa definen las capa- cidades para expresarse oral- La metodología elaborada en el marco de la educación en mente que pueden adquirirse con la utilización de la radio e, medios parece incluso permitir la inclinación de la sociedad incluso, se propone fomentar de la información hacia una sociedad del conocimiento, como defiende la UNESCO. En Francia, se necesitaría unir el interés por la lectura a través de la lectura de la prensa. La educación en los medios las fuerzas dispersas en función de los soportes mediáticos y orientarse más hacia la educación en medios que al dominio adquiere pleno derecho y entidad en la sección sexta titulada «competencias sociales y cívi- técnico de los aparatos. cas» que indica que «los alum- nos deberán ser capaces de juz- gar y tendrán espíritu crítico, lo que supone ser educados en los las programaciones oficiales, ya que, a lo largo de un medios y tener conciencia de su lugar y de su influencia estudio de los textos, los documentalistas del CLEMI en la sociedad». han podido señalar más de una centena de referencias a la educación de los medios en el seno de disciplinas 4. Un entorno positivo como el francés, la historia, la geografía, las lenguas, Si nos atenemos a las cifras, el panorama de la las artes plásticas : trabajos sobre las portadas de educación en medios es muy positivo. Una gran ope- prensa, reflexiones sobre temas mediáticos, análisis de ración de visibilidad como la «Semana de la prensa y publicidad, análisis de imágenes desde todos los ángu- de los medios en la escuela», coordinada por el CLE- los, reflexión sobre las noticias en los países europeos, MI, confirma año tras año, después de 17 convocato- información y opinión rias, el atractivo que ejerce sobre los profesores y los Esta presencia se constata desde la escuela mater- alumnos. Concebida como una gran operación de nal (2 a 6 años) donde, por ejemplo, se le pregunta a complementariedad (2000) Structure and mechanism of the aberrant ba3-cytochrome c oxidase from Thermus thermophilus. EMBO J 19: 1766–1776. 9. Hunsicker-Wang LM, Pacoma RL, Chen Y, Fee JA, Stout CD (2005) A novel cryoprotection scheme for enhancing the diffraction of crystals of recombinant cytochrome ba3 oxidase from Thermus thermophilus. Acta Crystallogr D Biol Crystallogr 61: 340–343. 10. Buschmann S, Warkentin E, Xie H, Langer JD, Ermler U, et al. (2010) The structure of cbb3 cytochrome oxidase provides insights into proton pumping. Science 329: 327–330. 11. Fee JA, Case DA, Noodleman L (2008) Toward a chemical mechanism of proton pumping by the B-type cytochrome c oxidases: application of density functional theory to cytochrome ba3 of Thermus thermophilus. J Am Chem Soc 130: 15002–15021. 12. Chang HY, Hemp J, Chen Y, Fee JA, Gennis RB (2009) The cytochrome ba3 oxygen reductase from Thermus thermophilus uses a single input channel for proton delivery to the active site and for proton pumping. Proc Natl Acad Sci U S A 106: 16169–16173. 13. Luna VM, Chen Y, Fee JA, Stout CD (2008) Crystallographic studies of Xe and Kr binding within the large internal cavity of cytochrome ba3 from Thermus thermophilus: structural analysis and role of oxygen transport channels in the heme-Cu oxidases. Biochemistry 47: 4657–4665. PLoS ONE | www.plosone.org 11 July 2011 | Volume 6 | Issue 7 | e22348 The 1.8 Å Structure of ba3Oxidase 27. Vogeley L, Sineshchekov OA, Trivedi VD, Sasaki J, Spudich JL, et al. (2004) Anabaena sensory rhodopsin: a photochromic color sensor at 2.0 A. Science 306: 1390–1393. 28. Cherezov V, Rosenbaum DM, Hanson MA, Rasmussen SG, Thian FS, et al. (2007) High-resolution crystal structure of an engineered human beta2adrenergic G protein-coupled receptor. Science 318: 1258–1265. 29. Jaakola VP, Griffith MT, Hanson MA, Cherezov V, Chien EY, et al. (2008) The 2.6 angstrom crystal structure of a human A2A adenosine receptor bound to an antagonist. Science 322: 1211–1217. 30. Wu B, Chien EY, Mol CD, Fenalti G, Liu W, et al. (2010) Structures of the CXCR4 chemokine GPCR with small-molecule and cyclic peptide antagonists. Science 330: 1066–1071. 31. Chien EY, Liu W, Zhao Q, Katritch V, Han GW, et al. (2010) Structure of the Human Dopamine D3 Receptor in Complex with a D2/D3 Selective Antagonist. Science 330: 1091–1095. 32. Hanson MA, Cherezov V, Griffith MT, Roth CB, Jaakola VP, et al. (2008) A specific cholesterol binding site is established by the 2.8 A structure of the human beta2-adrenergic receptor. Structure 16: 897–905. 33. Cherezov V, Liu W, Derrick JP, Luan B, Aksimentiev A, et al. (2008) In meso crystal structure and docking simulations suggest an alternative proteoglycan binding site in the OpcA outer membrane adhesin. Proteins 71: 24–34. 34. Cherezov V, Clogston J, Papiz MZ, Caffrey M (2006) Room to move: crystallizing membrane proteins in swollen lipidic mesophases. J Mol Biol 357: 1605–1618. 35. Caffrey M (2009) Crystallizing membrane proteins for structure determination: use of lipidic mesophases. Annu Rev Biophys 38: 29–51. 36. Loh HH, Law PY (1980) The role of membrane lipids in receptor mechanisms. Annu Rev Pharmacol Toxicol 20: 201–234. 37. Lee AG (2004) How lipids affect the activities of integral membrane proteins. Biochim Biophys Acta 1666: 62–87. 38. Robinson NC (1993) Functional binding of cardiolipin to cytochrome c oxidase. J Bioenerg Biomembr 25: 153–163. 39. Qin L, Sharpe MA, Garavito RM, Ferguson-Miller S (2007) Conserved lipidbinding sites in membrane proteins: a focus on cytochrome c oxidase. Curr Opin Struct Biol 17: 444–450. 40. Sedlak E, Panda M, Dale MP, Weintraub ST, Robinson NC (2006) Photolabeling of cardiolipin binding subunits within bovine heart cytochrome c oxidase. Biochemistry 45: 746–754. 41. Hunte C, Richers S (2008) Lipids and membrane protein structures. Curr Opin Struct Biol 18: 406–411. 42. Reichow SL, Gonen T (2009) Lipid-protein interactions probed by electron crystallography. Curr Opin Struct Biol 19: 560–565. 43. Yang YL, Yang FL, Jao SC, Chen MY, Tsay SS, et al. (2006) Structural elucidation of phosphoglycolipids from strains of the bacterial thermophiles Thermus and Meiothermus. J Lipid Res 47: 1823–1832. 44. Belrhali H, Nollert P, Royant A, Menzel C, Rosenbusch JP, et al. (1999) Protein, lipid and water organization in bacteriorhodopsin crystals: a molecular view of the purple membrane at 1.9 A resolution. Structure 7: 909–917. 45. Long SB, Tao X, Campbell EB, MacKinnon R (2007) Atomic structure of a voltage-dependent K+ channel in a lipid membrane-like environment. Nature 450: 376–382. 46. Gonen T, Cheng Y, Sliz P, Hiroaki Y, Fujiyoshi Y, et al. (2005) Lipid-protein interactions in double-layered two-dimensional AQP0 crystals. Nature 438: 633–638. 47. Seelig A, Seelig J (1977) Effect of single cis double bound on the structure of a phospholipid bilayer. Biochemistry 16: 45–50. 48. Lomize MA, Lomize AL, Pogozheva ID, Mosberg HI (2006) OPM: orientations of proteins in membranes database. Bioinformatics 22: 623–625. 49. Hoyrup P, Callisen TH, Jensen MO, Halperin A, Mouritsen OG (2004) Lipid protrusions, membrane softness, and enzymatic activity. Phys Chem Chem Phys 6: 1608–1615. PLoS ONE | www.plosone.org 50. Lee HJ, Svahn E, Swanson JM, Lepp H, Voth GA et al (2010) Intricate Role of Water in Proton Transport through Cytochrome c Oxidase. J Am Chem Soc 132: 16225–16239. 51. Aoyama H, Muramoto K, Shinzawa-Itoh K, Hirata K, Yamashita E, et al. (2009) A peroxide bridge between Fe and Cu ions in the O2 reduction site of fully oxidized cytochrome c oxidase could suppress the proton pump. Proc Natl Acad Sci U S A 106: 2165–2169. 52. Schmidt B, McCracken J, Ferguson-Miller S (2003) A discrete water exit pathway in the membrane protein cytochrome c oxidase, Proc Natl Acad Sci U S A 100: 15539–15542. 53. Einarsdottir O, Choc MG, Weldon S, Caughey WS (1988) The site and mechanism of dioxygen reduction in bovine heart cytochrome c oxidase. J Biol Chem 263: 13641–13654. 54. Agmon N (1995) The Grotthuss mechanism. Chem Phys Lett 244: 456– 462. 55. Salomonsson L, Lee A, Gennis RB, Brzezinski P (2004) A single amino-acid lid renders a gas-tight compartment within a membrane-bound transporter. Proc Natl Acad Sci U S A 101: 11617–11621. 56. Yin H, Feng G, Clore GM, Hummer G, Rasaiah JC (2010) Water in the polar and nonpolar cavities of the protein Interleukin-1-beta. J Phys Chem B 114: 16290–16297. 57. Eisenberger P, Shulman RG, Brown GS, Ogawa S (1976) Structure-function relations in hemoglobin as determined by x-ray absorption spectroscopy. Proc Natl Acad Sci U S A 73: 491–495. 58. Chishiro T, Shimazaki Y, Tani F, Tachi Y, Naruta Y, et al. (2003) Isolation and crystal structure of a peroxo-brodged heme-copper complex. Ang Chem Int Ed 42: 2788– 2791. 59. Ostermeier C, Harrenga A, Ermler U, Michel H (1997) Structure at 2.7 A resolution of the Paracoccus denitrificans two-subunit cytochrome c oxidase complexed with an antibody FV fragment. Proc Natl Acad Sci U S A 94: 10547–10553. 60. Kaila VRI, Oksanen E, Goldman A, Bloch D, Verkhovsky MI, et al. (2011) A combined quantum chemical and crystallographic study on the oxidized binuclear center of cytochrome c oxidase. Biochim Biophys Acta 1807: 769–778. 61. Sakaguchi M, Shinzawa-Itoh K, Yoshikawa S, Ogura T (2010) A resonance Raman band assignable to the O-O stretching mode in the resting oxidized state of bovine heart cytochrome c oxidise. J Bioenerg Biomembr 42: 241–243. 62. Chance B, Saronio C, Waring A, Leigh Jr. JS (1978) Cytochrome c-cytochrome oxidase interactions at subzero temperatures. Biochim. Biophys. Acta 503: 37–55. 63. Cheng A, Hummel B, Qiu H, Caffrey M (1998) A simple mechanical mixer for small viscous lipid-containing samples. Chem Phys Lipids 95: 11–21. 64. Cherezov V, Peddi A, Muthusubramaniam L, Zheng YF, Caffrey M (2004) A robotic system for crystallizing membrane and soluble proteins in lipidic mesophases. Acta Crystallogr D Biol Crystallogr 60: 1795–1807. 65. Minor W, Cymborowski M, Otwinowski Z, Chruszcz M (2006) HKL-3000: the integration of data reduction and structure solution - from diffraction images to an initial model in minutes. Acta Crystallogr D Biol Crystallogr 62: 859–866. 66. McCoy AJ, Grosse-Kunstleve RW, Adams PD, Winn MD, Storoni LC, et al. (2007) Phaser crystallographic software. J Appl Crystallogr 40: 658–674. 67. McRee DE (2004) Differential evolution for protein crystallographic optimizations. Acta Crystallogr D Biol Crystallogr 60: 2276–2279. 68. Unno M, Chen H, Kusama S, Shaik S, Ikeda-Saito M (2007) Structural characterization of the fleeting ferric peroxo species in myoglobin, Experiment and theory. J Am Chem Soc 129: 13394–13395. 69. Kuhnel K, Derat E, Terner J, Shaik S, Schlicting I (2007) Structure and quantum chemical characterization of chloroperoxidase compound 0, a common reaction intermediate of diverse heme enzymes. Proc Natl Acad Sci U S A 104: 99–104. 12 July 2011 | Volume 6 | Issue 7 | e22348
A Tunable Checkpointing Algorithm for Distributed Mobile Applications A Tunable Checkpointing Algorithm For Distributed Mobile Applications
RECENT ACTIVITIES
Autor
Documento similar

A Tunable Checkpointing Algorithm for Distributed Mobile Applications

Livre