Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Elastic{ON} 2018 - Making the business case for Elastic Stack

Elastic Co
March 01, 2018

Elastic{ON} 2018 - Making the business case for Elastic Stack

Elastic Co

March 01, 2018
Tweet

More Decks by Elastic Co

Other Decks in Technology

Transcript

  1. Elastic
    27-Feb-2018
    Twitter: @cjgeode
    How to make the business case for the
    Elastic Stack
    Catherine Johnson, Director of Product

    View full-size slide

  2. Who am I?
    Consulting -> Engineering -> Solution Architect -> Product
    Both business and technical background
    I am better with words than diagrams

    View full-size slide

  3. Why am I talking about this?
    “I want to put my logs into Elasticsearch” – WHY?
    Data dumps everywhere….
    I rarely see total cost considered

    View full-size slide

  4. TLDR….
    Cost is about more than license and hardware cost of a specific solution
    Risk is what you are really managing
    Added value is an important part of the calculation, but hard to determine
    You will always be challenged in comparing “apples to purple”
    What questions should you be asking yourself and the business?

    View full-size slide

  5. How strategic is my data?

    View full-size slide

  6. • Think about how this data will be used and by who
    • Will this identify gaps or opportunities for the business?
    • Will this help the efficiency of the business?
    • Will it make the business more competitive?
    • Think not just about how the data you are putting into Elastic is being used, but what you
    might want to do with that data in the next 1-2 years.
    • Do you envision new consumers of this data? How would it look different?
    • What other data might become important??
    What’s the value of the data?

    View full-size slide

  7. • Elastic is about putting data into a form that is easy to use and search
    • You want to make sure you keep what you want to search, and not just everything
    • You will hear people talk about optimizing field mappings… will you ever do a text
    search on a GUID? NOPE.
    • Don’t create a “data dump”
    Just because you *can* keep it…

    View full-size slide

  8. • Think about how you USE the data
    • Dashboards? Alerts? Analytics?
    • Fast? Slow? Really slow????
    • Are the users machines or humans?
    • What are your ”users” expectations?
    • What happens if the information is not available?
    • What other information is useful in context with this data?
    How is the information used?

    View full-size slide

  9. How will I get INFORMATION out of
    it?

    View full-size slide

  10. • Data stored neatly and efficiently on disk to make the most out of storage isn’t making
    the data USABLE.
    • Who is consuming it? What are they looking for? How easy will it be for them to
    decide to analyze the data in a new way? Add new types of data?
    • If you are ingesting 10TB a day, how are you planning on getting “information” out of
    this?
    • The US Library of Congress is 10TB. Simply making this data “available” doesn’t
    make it useful. How will you identify patterns and send meaningful alerts?
    To USE data strategically, you need to make is USE-able

    View full-size slide

  11. • When you need to solve a problem…..
    • How/when do you identify it? How long did this take? (MTTDetection)

    Are you looking at HUMAN TIME or MACHINE TIME?

    When does it cross that barrier? What is the additional impact?
    • Once you find it, how do you resolve it? (MTTRecovery)
    • The WHOLE THING matters, in fact detection REALLY matters
    Collecting 100TB isn’t enough

    View full-size slide

  12. • To consume large amounts of data, you will need strong algorithms available
    • Depending on your solution, this could be a secondary product or cost, typically
    requiring a separate skillset and separate hardware
    • What is the availability of the secondary solution? What happens when it goes
    down?
    • What is the hardware profile required to meet your SLAs?
    • How much hardware will be needed to meet your availability requirements?
    The full solution consideration

    View full-size slide

  13. How much will an outage cost?

    View full-size slide

  14. System ”value”: A million dollars in revenue a
    year
    $1m annually in revenue = $115 per HOUR
    “But it’s a million dollars!”

    View full-size slide

  15. Who uses it and HOW?
    How does it impact your overall resiliency?
    Machines are pickier than humans
    What part of the business do those systems
    support? What happens if THOSE are down?

    View full-size slide

  16. How does an outage impact brand reputation?
    This is the fuzzy part
    Bounding box (low and high), but the in
    between is unknown

    View full-size slide

  17. What is your downtime SLA?
    Your infrastructure “information” system cannot
    be the weakest link

    View full-size slide

  18. What is the cost of the supported system
    going down?
    Consider what the total risk is
    1. Does it support revenue functions for the company?
    2. Does the supported system play a role in regulatory compliance? What is the
    penalty for an outage?
    3. How does it impact resiliency?
    4. How would this impact your brand reputation?
    Past events can be used as barometers

    View full-size slide

  19. Now that you have this information

    View full-size slide

  20. • Think about those needs in choosing a data platform to host this information
    • Every solution will have a unique profile in terms of:
    • Hardware
    • Implementation
    • Ancillary solutions required
    • Then there is…..
    Now that you understand the size of the box

    View full-size slide

  21. • What new streams of information can you anticipate?
    • How will your volumes change?
    • How does your choice impact your future data vision?
    • Do you HAVE a future data vision?
    • Think more holistically
    • Or maybe you need a point solution?
    The future?

    View full-size slide

  22. Acceleration

    View full-size slide

  23. • Open source does not mean FREE
    • Is there a business advantage to having Elastic experts on staff?
    • If you go open source or with an “alternate” implementation, consider
    • The cost of other systems needed to support the basics (security)
    • The cost of other systems needed to do dedicated analysis of data
    • Correlation of data that would typically be in separate systems
    • Data is in one place, one interface, not multiples
    How do we help you reduce your time to market

    View full-size slide

  24. And the answer is…….

    View full-size slide

  25. • Come up with the hard numbers you can – no one will have all of them
    • Understand the real value of the data and WHY you are keeping it
    • Work with all of your vendors to figure out the *real* cost based on your environment
    • What is truly needed to meet your requirements?
    • Anticipate growth based on what you have seen the last 1-2 years
    • Software license + hardware profile of the complete solution
    • Does this number make sense compared to the value of the data?
    There is no simple answer….

    View full-size slide

  26. 26
    More Questions?
    Visit us at the AMA

    View full-size slide

  27. www.elastic.co

    View full-size slide