Cloudera

Everything you read in the positive reviews is true. Great leadership team, fantastic rank-and-file, huge market opportunity, meaningful technology that adds genuine knowledge and value to society at large.

Cons will be the focus of this review. Cloudera is at a critical juncture. The following is written out of sincere hope that Cloudera can improve. Potential new Clouderans, you have an opportunity to contribute to the solution; don't add to the problem.

(1) Problematic engineering middle management.

There are a few poisonous apples who are dragging down their teams and the company culture. Good managers are not being praised and rewarded/promoted. Well-meaning mediocre managers are not getting opportunities to train/improve. Manager hiring is starting to slide. Poisonous apples and mediocre managers start to bring in people that lead to nepotism and empire-building. The leadership team have to spend too much energy playing adjudicator and diplomat regarding middle-management petty fights or rank-and-file pointing out dubious decisions.

(2) Diluted Cloudera culture.

Cloudera had a unique culture that was once the envy of other companies. It was respectful and truly open, where good ideas could get embraced regardless of origin and rank. Several things diluted the culture.

• Open discussion somehow evolved into grandstanding. When that was rightly discouraged, there was a collateral chilling effect on positive discussion.
• Long-tenured employees emphasizing their experience lead to perceptions of diva-like behavior, however unintentionally. This compounds the chilling effect on open discussion. It also encourages problematic middle management to hide behind seasoned employees while eschewing hard calls and accountability.
• Parts of management becoming infatuated with and trying to copy cultures at other companies, without critically assessing Cloudera's unique strengths and other cultures' well-known weaknesses. This is disheartening to the rank-and-file because their peers outside of Cloudera are already panning the cultures that management tries to copy.
• Personal heroism and long hours glorified. This was not a problem during the necessarily chaotic startup phase. It has become a problem because heroism masks the need to address structural deficiencies; heroism should be rare and exceptional, not frequent and expected. Engineering morale and talent retention both suffered. There is a cultural drag as the business aims for efficiency and predictability in preparation for IPO.

(3) Open source becoming a burden.

This topic receives little visible discussion given Cloudera's sincere, thoughtful, and long-standing commitment to open source. There are a number of issues beyond the common open source detractor material in the trade press.

• Misalignment between the open source community and Cloudera's business. Work that will earn committer status is not necessarily work that is highest priority for Cloudera. If engineering career development within Cloudera is solely tied to committer status, and people without that status are considered second-class citizens, it introduces considerable incentive and misalignment issues.
• Engineering overhead and quality. This is an issue already well-felt by customers and well-appreciated by the leadership team. Cloudera's priority should be to deliver a polished product that customers love. A path to committer status is to deliver a major new feature. Polishing the product involves controlling the product surface area, but building new features will expand it. Over time, the open source bias in favor of new features accumulates into huge engineering overheads: merging changes made outside of Cloudera, making different components mutually compatible, bringing external code to high quality. Cloudera engineering should be Cloudera first. There is inherent misalignment if open source committer status is the sole measure of contribution and respect.
• Inherent shortcomings in integration, usability, and time to value. Big data use cases are expanding and diversifying. A rational approach is to solve new problems with existing systems, and when necessary, extend or re-architect on some fixed foundation. Open source incentives lead to an arguably irrational outcome: Whenever opportunity arises, develop an entirely new system, even if existing systems can sufficiently though imperfectly solve the problem. Do so because a new system leads to new project PMC and committer rosters, and hence additional opportunities for self-advancement. This outcome creates a huge and unnecessary burden for both Cloudera and its customers. Cloudera has to spend an extraordinary amount of effort juggling an out-of-hand product suite (~15 major CDH components and counting, each component with irrationally growing product surface area). Customers spend an extraordinary amount of effort learning the product and getting it to work. The product should be as close to "turn-key" as possible. Customers' efforts should focus on the data analysis itself, and the data tools should become invisible. To paraphrase a Cloudera co-founder, the best minds of my generation are spending time assembling glorified calculators; that sucks.

(4) Talent drain, the bad kind.

Rank-and-file Clouderans are absolutely amazing. They have moved mountains under extraordinary circumstances to get Cloudera to where it is today. It's one thing if talent leaves positively to forge their own path and burnish Cloudera's external credentials. It's another if talent leaves because of mind-numbing inflexibility in career development or compensation. It's even worse if talent leaves in groups to avoid poisonous apples in middle-management. Cloudera's big data leader position means that no matter where departing talent lands, it lands on potential competitors.

Speaking as a shareholder...

Immediate action: Identify and remove overpaid, poisonous apples in engineering middle management. This will have an immediate impact on engineering morale, with the side benefit of decreasing cash burn and stock dilution as Cloudera prepares for a potential IPO.

Allow anonymous questions during large meetings; gather regular anonymous feedback. Cloudera's atmosphere is no longer one where people feel comfortable enough to voice disagreements in public. But there remains a strong sense of ownership among the rank-and-file that anonymous channels can help the leadership see how things truly are, not just how things are postured by middle management.

Harder topics:

• Adjust culture and values: Don't fetishize other tech companies. Cloudera will not reach its full potential by repeating the mistakes of Google, Facebook, Twitter, or Uber. Cloudera's business and technology problems are, in some aspects, way harder, and it's up to Cloudera to find the Cloudera way.
• Foster structure and discipline without hampering initiative and innovation, for both engineering and the business at large. Important as Cloudera aims to scale further.
• Address concerns about compensation and career growth. People are air. No people, no air; the company suffocates.
• Create enough business and mental space to tackle hard problems around open source: how to get its best while minimizing its burden, how to make the technology useful and coherent, how to monetize it, and how to help customers monetize it.

Summary:

Cloudera is well-positioned to have a successful exit. Cloudera can determine whether it will be a middling success that will come and go, or it becomes the industry standard-bearer that it still has the potential to be, or whether all Clouderans will look back many years from now and smile with pride.