NOSQL stores offer good solutions for webscale apps but are suffering from fundamental limitations in the storage stack, hindering their performance in many ways like having highly unpredictable performance under load. Acunu has re-engineered the storage stack for the purpose of big data in what we believe to be the "most ambitious storage project undertaken since ZFS". Acunu is showcasing how it makes Apache Cassandra dramatically faster and more predictable while bringing unique one-click data cloning that will transform the lives of developers during the dev-test cycle.
Big Data applications: Scale + Performance + Low Cost (at least lower than Oracle). Pick any three. Individual data items are quite likely of low value. Losing (some of) them may even be acceptable. But the value comes from the scale, and intelligence (and competitive advantage) will come from the real-time analytics you can apply to the aggregate.\n
Big Data applications: Scale + Performance + Low Cost (at least lower than Oracle). Pick any three. Individual data items are quite likely of low value. Losing (some of) them may even be acceptable. But the value comes from the scale, and intelligence (and competitive advantage) will come from the real-time analytics you can apply to the aggregate.\n
Old\nOracle, MySQL, etc; on top of legacy FS, volume mgr, RAID, hardware\nNew\nNoSQL (to get horizontal scalability)\nHardware evolving fast - but not evenly, with SSD or large SATA disks\nHowever the stuff in the middle hasn’t changed much and b\n\n\n
"Big Data" applications are currently being built on top of a software stack - file systems,\nRAID, etc - that dates back decades. This legacy stack makes assumptions about the\ncapacity and performance of storage hardware which are no longer valid. Applications\nstruggle to take full advantage of modern hardware while they are locked into this envi-\nronment. Acunu has created a new storage back-end to take full advantage of modern\nlow cost hardware: multi-core CPUs, SSDs and low cost, high capacity SATA disks. Our\nfirst product combines Apache Cassandra, one of the leading NOSQL databases with\nour new back-end, delivering consistent and predictable high performance, fast tuning-\nfree deployment and easy management. For customers, this means sustained high\nthroughput with no garbage collection pauses or deployment surprises and performance\nup to 100X that of vanilla Cassandra.\n
"Big Data" applications are currently being built on top of a software stack - file systems,\nRAID, etc - that dates back decades. This legacy stack makes assumptions about the\ncapacity and performance of storage hardware which are no longer valid. Applications\nstruggle to take full advantage of modern hardware while they are locked into this envi-\nronment. Acunu has created a new storage back-end to take full advantage of modern\nlow cost hardware: multi-core CPUs, SSDs and low cost, high capacity SATA disks. Our\nfirst product combines Apache Cassandra, one of the leading NOSQL databases with\nour new back-end, delivering consistent and predictable high performance, fast tuning-\nfree deployment and easy management. For customers, this means sustained high\nthroughput with no garbage collection pauses or deployment surprises and performance\nup to 100X that of vanilla Cassandra.\n
"Big Data" applications are currently being built on top of a software stack - file systems,\nRAID, etc - that dates back decades. This legacy stack makes assumptions about the\ncapacity and performance of storage hardware which are no longer valid. Applications\nstruggle to take full advantage of modern hardware while they are locked into this envi-\nronment. Acunu has created a new storage back-end to take full advantage of modern\nlow cost hardware: multi-core CPUs, SSDs and low cost, high capacity SATA disks. Our\nfirst product combines Apache Cassandra, one of the leading NOSQL databases with\nour new back-end, delivering consistent and predictable high performance, fast tuning-\nfree deployment and easy management. For customers, this means sustained high\nthroughput with no garbage collection pauses or deployment surprises and performance\nup to 100X that of vanilla Cassandra.\n
"Big Data" applications are currently being built on top of a software stack - file systems,\nRAID, etc - that dates back decades. This legacy stack makes assumptions about the\ncapacity and performance of storage hardware which are no longer valid. Applications\nstruggle to take full advantage of modern hardware while they are locked into this envi-\nronment. Acunu has created a new storage back-end to take full advantage of modern\nlow cost hardware: multi-core CPUs, SSDs and low cost, high capacity SATA disks. Our\nfirst product combines Apache Cassandra, one of the leading NOSQL databases with\nour new back-end, delivering consistent and predictable high performance, fast tuning-\nfree deployment and easy management. For customers, this means sustained high\nthroughput with no garbage collection pauses or deployment surprises and performance\nup to 100X that of vanilla Cassandra.\n
"Big Data" applications are currently being built on top of a software stack - file systems,\nRAID, etc - that dates back decades. This legacy stack makes assumptions about the\ncapacity and performance of storage hardware which are no longer valid. Applications\nstruggle to take full advantage of modern hardware while they are locked into this envi-\nronment. Acunu has created a new storage back-end to take full advantage of modern\nlow cost hardware: multi-core CPUs, SSDs and low cost, high capacity SATA disks. Our\nfirst product combines Apache Cassandra, one of the leading NOSQL databases with\nour new back-end, delivering consistent and predictable high performance, fast tuning-\nfree deployment and easy management. For customers, this means sustained high\nthroughput with no garbage collection pauses or deployment surprises and performance\nup to 100X that of vanilla Cassandra.\n
"Big Data" applications are currently being built on top of a software stack - file systems,\nRAID, etc - that dates back decades. This legacy stack makes assumptions about the\ncapacity and performance of storage hardware which are no longer valid. Applications\nstruggle to take full advantage of modern hardware while they are locked into this envi-\nronment. Acunu has created a new storage back-end to take full advantage of modern\nlow cost hardware: multi-core CPUs, SSDs and low cost, high capacity SATA disks. Our\nfirst product combines Apache Cassandra, one of the leading NOSQL databases with\nour new back-end, delivering consistent and predictable high performance, fast tuning-\nfree deployment and easy management. For customers, this means sustained high\nthroughput with no garbage collection pauses or deployment surprises and performance\nup to 100X that of vanilla Cassandra.\n
"Big Data" applications are currently being built on top of a software stack - file systems,\nRAID, etc - that dates back decades. This legacy stack makes assumptions about the\ncapacity and performance of storage hardware which are no longer valid. Applications\nstruggle to take full advantage of modern hardware while they are locked into this envi-\nronment. Acunu has created a new storage back-end to take full advantage of modern\nlow cost hardware: multi-core CPUs, SSDs and low cost, high capacity SATA disks. Our\nfirst product combines Apache Cassandra, one of the leading NOSQL databases with\nour new back-end, delivering consistent and predictable high performance, fast tuning-\nfree deployment and easy management. For customers, this means sustained high\nthroughput with no garbage collection pauses or deployment surprises and performance\nup to 100X that of vanilla Cassandra.\n
"Big Data" applications are currently being built on top of a software stack - file systems,\nRAID, etc - that dates back decades. This legacy stack makes assumptions about the\ncapacity and performance of storage hardware which are no longer valid. Applications\nstruggle to take full advantage of modern hardware while they are locked into this envi-\nronment. Acunu has created a new storage back-end to take full advantage of modern\nlow cost hardware: multi-core CPUs, SSDs and low cost, high capacity SATA disks. Our\nfirst product combines Apache Cassandra, one of the leading NOSQL databases with\nour new back-end, delivering consistent and predictable high performance, fast tuning-\nfree deployment and easy management. For customers, this means sustained high\nthroughput with no garbage collection pauses or deployment surprises and performance\nup to 100X that of vanilla Cassandra.\n
"Big Data" applications are currently being built on top of a software stack - file systems,\nRAID, etc - that dates back decades. This legacy stack makes assumptions about the\ncapacity and performance of storage hardware which are no longer valid. Applications\nstruggle to take full advantage of modern hardware while they are locked into this envi-\nronment. Acunu has created a new storage back-end to take full advantage of modern\nlow cost hardware: multi-core CPUs, SSDs and low cost, high capacity SATA disks. Our\nfirst product combines Apache Cassandra, one of the leading NOSQL databases with\nour new back-end, delivering consistent and predictable high performance, fast tuning-\nfree deployment and easy management. For customers, this means sustained high\nthroughput with no garbage collection pauses or deployment surprises and performance\nup to 100X that of vanilla Cassandra.\n
* performance is not just about speed, but also reliability + predictability ... but a Volvo isn’t as sexy as a McLaren F1!\n* note the specific version tree, showing live / dev branches\n* monitoring goes from the cluster down to iostat\n* ... and includes management: easy cluster deployment\n\n
* performance is not just about speed, but also reliability + predictability ... but a Volvo isn’t as sexy as a McLaren F1!\n* note the specific version tree, showing live / dev branches\n* monitoring goes from the cluster down to iostat\n* ... and includes management: easy cluster deployment\n\n
* performance is not just about speed, but also reliability + predictability ... but a Volvo isn’t as sexy as a McLaren F1!\n* note the specific version tree, showing live / dev branches\n* monitoring goes from the cluster down to iostat\n* ... and includes management: easy cluster deployment\n\n
* performance is not just about speed, but also reliability + predictability ... but a Volvo isn’t as sexy as a McLaren F1!\n* note the specific version tree, showing live / dev branches\n* monitoring goes from the cluster down to iostat\n* ... and includes management: easy cluster deployment\n\n
* performance is not just about speed, but also reliability + predictability ... but a Volvo isn’t as sexy as a McLaren F1!\n* note the specific version tree, showing live / dev branches\n* monitoring goes from the cluster down to iostat\n* ... and includes management: easy cluster deployment\n\n
* performance is not just about speed, but also reliability + predictability ... but a Volvo isn’t as sexy as a McLaren F1!\n* note the specific version tree, showing live / dev branches\n* monitoring goes from the cluster down to iostat\n* ... and includes management: easy cluster deployment\n\n
* performance is not just about speed, but also reliability + predictability ... but a Volvo isn’t as sexy as a McLaren F1!\n* note the specific version tree, showing live / dev branches\n* monitoring goes from the cluster down to iostat\n* ... and includes management: easy cluster deployment\n\n
* performance is not just about speed, but also reliability + predictability ... but a Volvo isn’t as sexy as a McLaren F1!\n* note the specific version tree, showing live / dev branches\n* monitoring goes from the cluster down to iostat\n* ... and includes management: easy cluster deployment\n\n
1 box - 24GB RAM, 7 SATA disks.\n3 billion rows.\n8 byte keys.\n50 byte vales.\n1 column per row.\nBatch insert 100 at once.\n
cassandra node become inaccessible, node drop out\nit’s hard to design around latency that spike like this.\n
this is a behaviour after large number of writes and it takes time to recover. \n