Citation Context

... server sets up a TCP connection once with a remote server and sends many RPC over that connection. To handle many connections and RPCs simultaneously, the programs use SFS’s asynchronous RPC library =-=[15]-=-. 8.3 Experiment results This implementation provides Chord with high-performance for its operations. For example, on a PIII 733, the Chord server can process 10,300 lookup RPCs per second. We haven’t...

Citation Context

...ntally untrusted. Servers may crash without warning or leak information to third parties. This lack of trust is inherent in the utility model and is different from other cryptographic systems such as =-=[35]-=-. Only clients can be trusted with cleartext—all information that enters the infrastructure must be encrypted. However, rather than assuming that servers are passive repositories of information (such ...

Citation Context

...s [9, 6, 7, 13, 29, 34]. The use of content-hashes to securely link together different pieces of data is due to Merkle [18]; the use of public keys to authentically name data is due to the SFS system =-=[17]-=-. CFS adopts naming, authentication, and file system structure ideas from SFSRO [9], which implements a secure distributed readonly file system---that is, a file system in which files can be modified ...

Citation Context

...espite a potentiallysmalicious server, but it does this by placing trust in allsclient machines (unlike Farsite, which requires only thatseach user trust the client machine he is directly using).sSFS =-=[26]-=- constructs "self-certifying pathnames" bysembedding hashes of public keys into file names, thussdefending read-only data against compromised serverssor compromised networks.sA number of distributed s...

Citation Context

...ist a 7chunk index LBFS client TCP LBFS server chunk index xfs client NFS server Figure 4: Overview of the LBFS implementation. wider range of attacks. LBFS uses the security infrastructure from SFS =-=[16]-=-. Every server has a public key, which the client administrator specifies on the command line when mounting the server. In the future, we intend to embed public keys in pathnames as SFS does and to in...

Citation Context

...oblem: An initiator chooses a new tunnel successor only with 6 probability or when fails. 39Chapter 5 Implementation We have implemented Tarzan in C++ on Unix to validate our approach. SFS libraries =-=[17]-=- provide callback-style functionality for fast asynchronous I/O and automate marshalling data into the standard XDR wire representation. The core component of Tarzan is a stand-alone relay server whic...

Citation Context

...m a file system with a different set of logs, they create a new view block. A user names an Ivy file system with the content-hash key of the view block; this is essentially a self-certifying pathname =-=[23]-=-. Users creating a new file system must exchange public keys in advance by some out-of-band means. Once they know each other's public keys, one of them creates a view block and tells the other users t...

Citation Context

...ve enough context to make the necessary connections. For now, we merely note that almost every aspect of our design is (proudly) stolen from elsewhere, most notably from TRIAD [19], HIP [28], and SFS =-=[25]-=-. It is the synthesis of these various ideas into a coherent architecture that we claim as our contribution. The paper ends, in Section 8, with some speculations on DONA’s broader architectural implic...