企业存储类初创公司走红
Katherine Noyes | 2015-01-04 15:56
分享: [译文]
Thirty years ago, before software began “eating the world,” much of the information in industry was stored on spinning disks—those relatively slow but inexpensive devices reminiscent of record players. A handful of manufacturers competed in that business. The business of relational databases—the software typically used to structure the data as they were stored—was largely dominated by Oracle.
Industry hummed along that way for many years. Then it all changed, quickly.
First, Apple’s late CEO Steve Jobs refused to continue using spinning drives in the company’s iPod portable music player, choosing instead lighter, more energy-efficient solid-state drives. Given the tremendous volumes of the iPod that were sold at the time, the decision helped drive a movement toward flash memory in consumer devices, says John Vrionis, a partner at Lightspeed Venture Partners. Prices for flash began to fall dramatically. Large businesses and their data centers soon followed.
“Flash performs like memory, which is 100 times faster than spinning disks, but it’s almost as cheap as disk,” Vrionis says. “You get all the performance benefits of memory, but it costs almost the same as disk.”
Another contributing factor was a gigantic leap in the amount of data that average users were accessing on a given day, supported by the rapid rise of the Internet-connected mobile device. On a consumer level, anyone with a smartphone can relate to the trend; for businesses, the volumes began to get so big as to fall into the territory now known as “big data.”
A final factor was the storage industry’s move to software-defined storage, a kind of storage technique that uses cheap, distributed hardware (rather than massive, centralized boxes) to scale with demand easily and inexpensively. “Just as you want to buy whatever PC you want and then choose the operating system separately, people want to buy hardware and then get the storage software separately,” Vrionis says.
Put it all together, and what do you get? A storage market ripe for change and a boatload of startups vying to make it happen. “Huge tailwinds are giving all these startups a chance,” Vrionis says. “They’re unstoppable now.”
$1 billion per year
Vrionis’s firm is just one of many that have been betting heavily on enterprise storage startups—including contenders such as DataStax and PernixData, which recently drew fresh funding of $106 million and $35 million, respectively. (Vrionis himself serves on DataStax’s board of directors.)
A few other examples that have drawn investment attention in the past few months: Primary Data, which took in $10 million in Series B funding in February; Tintri, which garnered $75 million in Series E funding that same month; Pure Storage, which closed a $225 million Series F round in April; Nutanix, with a $140 million Series E in August; Amplidata, with a $10 million Series E round in September; Formation Data Systems, with a $24.2 million Series A that month; SolidFire, which closed an $82 million Series D round in October; SwiftStack, whose $16 million Series B round also took place that month; IzumoBase, which raised $1.4 million in November; Kaminario, whose $53 million Series E took place earlier this month; and DataGravity, which this month closed a $50 million Series C round.
In all, disclosed investment in enterprise-storage startups has amounted to about a billion dollars per year in each of the last few years, says Simon Robinson, a research vice president for storage and information management with 451 Research, which has been covering the space since 2000.
“You get these waves of startups,” Robinson says. “As soon as you think the industry is done with storage, along comes another wave. It’s been a constant cycle of innovation in the last decade.”
Typically, those waves of innovation have each focused on a particular feature, spawning numerous startups playing to a similar theme. “Some wouldn’t last; a couple would be acquired; one or two would do really well and go public,” Robinson says. “Then usually one of the big guys would step in and pay a substantial premium to acquire them.”
Data Domain, for instance, was acquired in 2009 by EMC EMC 1.00% , which snatched up Isilon the following year. Hewlett-Packard HPQ -0.25% bought 3PAR in 2010; Dell acquired Compellent Technologies in 2011. The list goes on.
“The thing to bear in mind is that you have five or six big vendors, virtually nobody in middle, and then a whole bunch of startups,” Robinson says. “We haven’t seen a major change in the cast of characters in the last decade.”
Dave Wright, SolidFire‘s founder and CEO, says his company grew 700% year over year in 2013 and 50% quarter over quarter this year. He attributes much of that to two “huge waves of change” affecting the industry.
The shift to flash and solid-state storage is one of those. “That’s very disruptive—a huge chunk of the storage market is going to move to flash over the next few years,” he says. (SolidFire offers all-flash systems with a focus on scale-out storage for data centers.)
At the same time, though, Wright downplays that trend as “a fairly short-term disruptive shift—everybody is doing that,” he says.
More interesting from Wright’s perspective is a second major shift: the transition within enterprises from traditional, so-called siloed data infrastructures to cloud-based ones. It’s a change that is affecting the entire industry, but one that “a lot of other startups in the storage space are not really attuned to,” he says. ‘That’s what we think is going to make us more successful in the long term: SolidFire is the only infrastructure purpose-built for this wave of cloud.”
‘Looking for simplicity’
Part of the challenge for storage startups today is to simplify what has become an inordinately complex and fragmented landscape, Robinson says. It’s the result of a lot of feature-focused research and development over the last decade.
“Big companies have lots of storage systems, each of which serves a particular purpose,” he says. “The challenge that organizations are having now is not just the data growth, but managing that environment, which may consist of seven or eight storage silos with huge complexity.”
Simplicity is a new goal, in other words. The trend has led to an industry preference for more all-encompassing approaches. “Converged” (which combines storage and compute in a single device) and “hyper-converged” (which combines storage, compute, networking, and virtualization in a single device) models reflect the shift. Robinson points to Nutanix and Formation Data Systems as notable contenders.
Meanwhile, the flood of apps—internal and customer-facing, and often mobile—continues unabated, increasing pressure on businesses to achieve better performance and speed from the databases and storage systems that underpin those applications.
“Every second of latency can cost you 10 percent of conversions, and it’s getting worse on mobile, where the network has its own built-in latency,” says Matt Murphy, a general partner at Kleiner Perkins Caufield & Byers. (KPCB invested in both DataStax and PernixData; Murphy serves on PernixData’s board.) “If you can cut off a second or half a second, there’s millions or hundred of millions of dollars in opportunity there in play. At all layers—storage, compute, network—everyone is trying to figure out how to get that extra edge.”
The question is which technologies and vendors will best deliver that edge. Lightspeed’sVrionis contends that newcomers to the scene have an advantage over incumbents, which are still “staring at yesterday’s problems and not able to take a blank sheet of canvas.” (One of those incumbents, Oracle, did not respond to Fortune‘s request for comment.)
Henry Baltazar, a senior analyst with Forrester Research, agrees. “The big guys are finally catching up, but they left the door open for too long,” he says. “Now it’s a battle.”
30年前,也就是软件还没有“蚕食这个世界”的时候,大量的行业信息都被储存在所谓的“旋转式磁盘”之中。这种存储速度相对较慢,但比较便宜的存储设备,很容易让人回忆起唱片机。在这个领域里竞争的制造商也为数不多。关系数据库(也就是用来构建存储数据的软件)产业,当时基本上是甲骨文公司的天下。
整个行业沿着这个方向发展了很多年,但很快一切都变了。
首先是苹果已故CEO史蒂夫•乔布斯不愿继续在iPod便携式音乐播放器中使用旋转式磁盘,转而使用质量更轻,也更节能的固态硬盘。光速创投合伙人约翰•弗里奥尼斯指出,由于当时iPod的销量极大,这项决定推动了闪存在消费电子设备中的使用。闪存的价格也开始显著下降。各大企业和他们的数据中心也很快跟上了这一潮流。
弗里奥尼斯表示:“闪存的性能与内存相似,但它比旋转式磁盘快100倍,而价格却像硬盘一样便宜。你可以既享受到内存的性能优势,同时享受到几乎像硬盘一样低廉的价格。”
另外一个推动因素,则是普通用户平均每天存取的数据量出现了巨大飞跃,而这主要归功于联网移动设备的快速兴起。在消费者层面上,任何人只要有一部智能手机,都与这种趋势相关。在企业界,数据量开始迅速增大,由此催生了一个现在被称为“大数据”的新兴领域。
最后一个因素是存储行业开始转向软件定义存储,这种存储技术使用了廉价、分散的硬件(而不是集中的大型箱式机)实现了规模效益,可以轻易地并且低成本地满足需求。弗里奥尼斯表示:“就像你可以买任何一台你喜欢的电脑,然后再单独选择操作系统一样,人们也想先买硬件,然后再单独买存储软件。”
将所有这些趋势聚集在一起,你能得到什么结论?那就是存储市场已经发展到变革的边缘,很多创业公司争相使之发生。弗里奥尼斯认为:“一股巨大的推动力正在给予所有这些创业公司一个机会,他们现在几乎是不可阻挡的。”
每年10亿美元
在企业存储领域下了重注的风投还有很多,弗里奥尼斯的公司只是其中之一。在这个领域中竞争的创业公司有DataStax和PernixData等,这两家公司最近刚刚分别拉到了1.06亿美金和3500万美元的投资。(弗里奥尼斯自己就是DataStax公司的董事会成员。)
除此之外,还有几家公司在过去几个月也吸引了大量投资。比如Primary Data公司在二月的B轮融资中融得1000万美元;Tintri公司也于二月在E轮融资中募得7500万美元;Pure Storage公司四月在F轮融资中筹得2.25亿美元;Nutanix公司在八月份的E轮融资中筹得1.4亿美元;Amplidata公司在九月份的E轮融资中筹得1000万美元;Formation Data Systems公司也于同月在A轮融资中筹得2420万美元;SolidFire公司在十月份的D轮融资中筹得8200万美元;SwiftStack公司在同月的B轮融资中筹得1600万美元;IzumoBase公司在11月融资140万美元;Kaminario公司在本月初的E轮融资中筹得5300万美元;DataGravity公司刚刚在本月的C轮融资中筹得5000万美元。
451 Research公司自从2000年起就开始关注企业存储领域,该公司的存储与信息管理研究副总裁西蒙•罗宾逊指出,总体来看,在过去几年里,向企业存储类创业公司已披露的投资,基本上每年都能达到10亿美元左右。
罗宾逊表示:“这类初创公司层出不穷,你刚觉得行业的这一波存储潮结束了,马上就会迎来另一波。过去十年,这个领域基本上一直都处于创新期。”
每一波的创新浪潮基本上都集中于某一特定功能,大量创业公司都在围绕一个类似的主题。罗宾逊指出:“有些创业公司不会持久,有一些会被收购,有一两家公司会做得非常出色,然后上市。然后某家大公司就会踏足进来,花高价收购它们。”
比如Data Domain公司就于2009年被EMC公司收购,就在第二年,EMC又收购了一家名叫Isilon的创业公司。惠普(Hewlett-Packard)则于2010年买下了3PAR公司;戴尔(Dell)于2011年收购了Compellent Technologies公司。类似的收购还有很多。
罗宾逊指出:“需要注意的是,这个领域大约有五六家大厂商,基本上没有人在中游,然后就是一大堆创业公司。在过去10年里,这张‘演员表’基本上没怎么大变。”
SolidFire公司创始人兼CEO戴夫•莱特表示,他的公司2013年的年增率达到700%,而每个季度都比上一季度增长50%。他认为,这主要归功于影响行业的两个“重大的变革浪潮”。
首先是向闪存和固态硬盘的转变。他表示:“它非常有颠覆性,在未来几年,存储市场将有相当一部分向闪存转型。”(SolidFire主要提供全闪存系统,同时重点提供数据中心的扩展存储。)
不过与此同时,莱特也看低这股趋势,认为它只是“一种较为短期的颠覆式转型——人人都在做它。”
从莱特的观点看,更有意思的则是第二个重大转型,也就是企业从传统的所谓孤立式数据架构向云架构的转变。这种转变正在影响整个行业,但是“很多存储类创业公司并没有真正适应这种转变。”他还表示:“这就是我们为什么认为我们在长期会更加成功,因为SolidFire是目前唯一专门为云计算浪潮构建的专用架构。”
寻找简单性
罗宾逊认为,如今存储类创业公司面临的挑战之一,是要简化过去那种非常复杂和碎片化的格局。它正是过去十年很多以功能为重点的研发所造成的结果。
他表示:“大公司有很多存储系统,其中每个系统都服务于一个特定的用途。企业所面临的挑战不仅仅是数据量的增长,还有如何管理这种环境,因为这些系统可能包括了七八个独立的存储系统,极其复杂。”
换句话说,简单性是一个新的目标。这种趋势使得整个行业开始青睐那些“万金油”式的方案。所谓的“融合”模式(将存储和计算整合到一台设备上)和“高度融度”模式(将存储、计算、网络和虚拟化整合到同一台设备上)的流行也反映了这种转变。罗宾逊指出,Nutanix和Formation Data Systems公司是这方面的两个值得注意的竞争者。
与此同时,应用软件的热潮(包括内部应用、面向顾客的应用,往往还有移动应用)依然没有减退,这也加大了企业的压力,迫使它们不得不提高数据库和存储系统的性能和速度。
Kleiner Perkins Caufield & Byers公司(该公司是DataStax和PernixData公司的投资方,其合伙人墨菲也是PernixData的董事会成员)的合伙人马特•墨菲指出:“每一秒钟的延迟都可能让你损失10%的客户,在移动端则更严重,因为移动网络本来就有延迟。如果你可以把延迟减少一秒或半秒,就拥有了可能赚几百万甚至上亿美元的机会。在存储、计算和网络等所有层面上,每家公司都在想方设法地获得这种额外优势。”
现在的问题是:哪些技术和哪些厂商能够最好地将这种优势带给客户?光速创投公司的弗里奥尼斯也认为,这个领域的新进者其实比守成者更有优势,因为“守成者们仍然盯着昨天的问题,不能毅然扬帆前进。”(其中的一个“守成者”甲骨文公司并没有回复《财富》的评论请求。)
福雷斯特研究公司的高级分析师亨利•巴尔塔扎尔也表示:“大公司最终会赶上来,但是他们给创业公司留了太长时间的空子。现在,战争开始了。”(财富中文网)
译者:朴成奎
相关阅读: