小公司颠覆存储业大野心：冲击百亿美元数据库大市场

    兹魏本早先效力于美国宇航局艾姆斯研究中心（NASA Ames Research Center），此前还创立过“蓝色马提尼软件公司”（Blue Martini Software）。

    兹魏本称：“大多数想要应用这种向外扩展技术的竞争对手都专注于分析应用。而我们认为更有价值的可能是支持实时应用，也就是同时在线的一组用户在同一时间读取、写入数据库。这是无人曾涉足过的可扩展性，也正是我们现在的主攻方向。”

“一家颠覆业界的公司”

    Mohr Davidow公司普通合伙人比尔•埃里克森称，Splice Machine在很多方面让那些“拥有50个博士和无数Java程序员”的大公司所能做的事普及化了。这家公司参与了对Splice Machine的A轮及B轮投资。

    他说：“现有这些数据密集型技术的一个问题是，如果企业拥有聪明绝顶的博士，这些技术就能大显神通，但一般企业的资源很有限，”而使用Splice Machine的产品后，“成本只有用Oracle这类产品的10%到25%。用户没必要购买所有的新软件和新工具。他们不仅能省下大笔的基础设施开销，而且不必抛弃过去25年来辛辛苦苦打造的一切。”

    克利夫兰称，Splice Machine的模式“极大地改变了数据库扩展的经济性。Splice Machine的一位客户反映，与其目前的数据库解决方案相比，他们现在的查询速度提高了三到七倍，同时成本却降低了75%。”

    他进一步表示，对InterWest Partners公司来说——它参与了Splice Machine的B轮投资——这家公司“正是我们深信将颠覆业界的典型企业。因为Splice Machine支持标准SQL，它有潜力通过提供低成本、高效率、向外扩展的数据库解决方案代替（或者说颠覆）传统的数据库供应商，而无需让企业重写现有应用。相反，要用NoSQL和NewSQL公司的解决方案，就必须重写现有应用。依靠Splice Machine，编写现有SQL应用的程序员可以无须具备任何专门技巧或培训就能轻松使用Splice Machine的解决方案。”

一个潜在的收购对象

    Pund-IT公司的金表示，关于Splice Machine是否能兑现承诺，以及RDBMS供应商是否会因此忧心忡忡，这两点还有待观察。

    他说：“从这家公司15家首先试用的客户众口一词的赞扬可以发现，它的确没有说大话。而传统供应商是不是该感到担忧就不好说了。很多曾一度想撼动RDBMS市场的供应商都铩羽而归，乏善可陈。”

    金表示，很多企业在支持必要的业务应用和流程的同时，还在设法节约成本。这就意味着Splice Machine将会成为很多公司青睐的对象。

    金说：“如果这家公司确实能交付好的产品，它肯定会蒸蒸日上。它甚至会成为传统RDBMS企业的收购目标，因为这些企业既希望获得更多大数据技术的魔力，同时又让自己对手的日子不好过。”（财富中文网）

    译者：清远

    Previously, Zweben worked at the NASA Ames Research Center; he also founded Blue Martini Software.

    "The majority of competitors that are trying to apply this scale-out technology are focused on analytical apps," Zweben said. "What we think may be more valuable is powering real-time apps, where a concurrent set of users are reading and writing to the database at the same time. That's scalability nobody has been addressing, and that's what we do."

'A category-disruptive company'

    In many ways, Splice Machine democratizes what huge companies "with 50 Ph.Ds and countless Java programmers" can do, said Bill Ericson, a general partner at Mohr Davidow, which participated in the company's Series A and Series B financing rounds.

    "One of the problems with these data-intensive technologies is, they work great if you have really smart Ph.Ds but the average enterprise has limited resources," Ericson said. With Splice Machine, "the costs are probably 10 to 25 percent of trying to do this with something like Oracle. Users don't have to buy all new software and all new tools, and not only do they save on the fundamental infrastructure, but they don't have to throw out the window everything they built over the last 25 years."

    Splice Machine's model "dramatically changes the economics of scaling databases," Cleveland said. "One of Splice Machine's customers had their queries speed up by three to seven times while reducing costs by over 75% compared to their current database solution."

    For InterWest Partners -- which participated in Splice Machine's Series B -- the company is "an example of investing in what we believe could be a category-disruptive company," he added. "Since Splice Machine supports standard SQL, it has the potential to replace -- disrupt -- the traditional database suppliers by offering a low-cost, high-performance, scale-out database solution without companies having to rewrite their existing applications, which is what the NoSQL and NewSQL companies require in order to use their solutions. With Splice Machine, the same programmers who wrote the existing SQL applications can easily and simply use Splice Machine without any specialized skills or training."

A potential acquisition target

    It remains to be seen whether Splice Machine can deliver on its promises and whether RDBMS vendors have reason to worry, Pund-IT's King said.

    "The answer to the first is found in the accolades of the company's 15 pilot customers," he said. "Whether established vendors should be worried is harder to say. Any number of alternative vendors have tried to shake up the RDBMS market without much success to show for it."

    Still, organizations are searching for ways to save money while supporting necessary business applications and processes, King said. Which means Splice Machine could be the subject of much interest.

    "If it truly can deliver the goods, the company could do very well," King said. "[It could] even become an acquisition target by a traditional RDBMS vendor that wishes to capture a bit more big data magic while making life difficult for its competitors."