Louis Imershein

Louis Imershein

As Vice President of Product at Permabit, Louis Imershein is responsible for product management and strategic planning. He has 25 years of technical leadership experience in product management, software development and technical support.

Prior to Permabit, Imershein was a Senior Product Marketing Manager for the Sun Microsystems Data Management Group. Before joining Sun, he was at SCO where his titles included Product Manager, Software Architect and Senior Technical Lead.

Imershein has a Bachelor’s degree in Biological Science from the University of California, Santa Cruz and has earned certificates in software design, device driver development, system administration, and software testing.

curata__xibQsEPYnpZqlhw.png

A New Class of Storage: Data Reduction meets 3D XPoint

| By: (55)

The Intel Optane DC P4800X looks like a typical PCIe NVMe SSD – with one key difference.  Instead of using traditional NAND flash, it uses the much-anticipated 3D XPoint non-volatile memory technology developed by Intel and Micron. As the folks over at anandtech.com note, “3D XPoint enables this SSD to deliver high throughput and lower access latency than any other NVMe SSD on the market.” The use of 3D XPoint…

Read more

Permabit VDO on a Linux Laptop – Great Performance and 5:1 Space Savings

| By: (55)

  I get asked about VDO performance all the time and I’ve written several posts about big systems where we’ve seen spectacular performance numbers including 8 GB/s throughput and 650,000 mixed random 4 KB IOPS.  But what about performance on smaller systems for developers?  How about a laptop? A couple weeks ago I installed VDO version 6 on my Lenovo X230 laptop running Red Hat Enterprise Linux 7.3 and here’s a…

Read more

curata__UOYLNvle4ThKgip.jpeg

Addressing Bandwidth Challenges in the Hybrid Cloud

| By: (55)

Any application infrastructure that relies on a single data center is only as safe as that data center’s physical resources and the competence of its staff.  Witness the recent S3 outage at Amazon. When you choose to deploy in a single public cloud, you are delegating infrastructure management to your provider. When you’re exclusively running in-house, private cloud infrastructure, you’re entrusting that management to your own organization.  Either way mistakes…

Read more

curata__511a36eb738072f5fe45a1ca1d561ce0.PNG

Transparent Data Reduction in the Private Cloud

| By: (55)

Tom Cook, Permabit’s CEO, recently discussed how to control private cloud infrastructure and software costs by utilizing data reduction software. Data reduction makes private cloud more affordable than ever. This is why, as you’re readying your private cloud deployment, you should be sure it has data reduction capabilities from the get-go. How should you implement data reduction as part of your private cloud deployment? If you’re implementing on Linux, the…

Read more

curata__BM0mQtBrwzT8R4R.jpeg

Effective use of data reduction in the Public Cloud

| By: (55)

  Permabit CEO, Tom Cook recently wrote about how data reduction technology can simplify the problems associated with provisioning adequate storage resources in the public cloud, while balancing performance and efficiency.  The good news is, taking advantage of data reduction software in the public cloud is easier than ever. For example,Permabit’s Virtual Disk Optimizer (VDO) is a pre-packaged software solution that installs and deploys in minutes on Red Hat Enterprise…

Read more

curata__0lq7FVrFBRdgESp.jpeg

Latest VDO Performance Results – High-end Performance Meets Software Defined Storage

| By: (55)

Today Permabit announced that our Virtual Data Optimizer (VDO), using HIOPS Compression, achieved 8 GB/s throughput when tested on the Samsung NVMe Reference Design platform from Samsung Electronics, Ltd.  This is astounding performance from a single node Intel server platform, particularly one that has no custom hardware acceleration feature for compression. As a follow on, I wanted to provide some background on why we wanted to collect these numbers, how we…

Read more

curata__s8dZYCNYdbPFJ5l.png

CalSoft Joins the Permabit Integration Partner Program

| By: (55)

Last week Permabit announced a reseller partnership with Calsoft, Inc, a leading IT services company specializing in storage, networking, virtualization and cloud business verticals.  What may not be obvious from that announcement is that CalSoft is not new to Permabit.  They’ve been offering consulting services to OEMs integrating Permabit products for over five years!  With this announcement, they are extending the availability of their integration services for Permabit-enabled data reduction…

Read more

curata__QzJx51Vcuhn3ONY.png

The Paris Accord: Data Reduction Impact on the Environment

| By: (55)

You’re not going to see the environmental impact of data centers as an election day issue.  It’s just not something that galvanizes the public because it’s so removed from everyday life.  But, with US election day just around the bend, in an article over on LinkedIn,  Lynn Renshaw of Peerless Business Intelligence is thinking about how our increasing appetite for data will impact the Paris Accords on climate change, and…

Read more

curata__AJLNA97OQToCJN4.png

How to Save Over $370,000 with Permabit VDO and Red Hat Ceph Storage

| By: (55)

Over the last couple of weeks, Permabit has seen a huge outpouring of interest in our Red Hat Storage partnership announcement. Much of the interest, to date, has been around VDO with Ceph. We’ve seen some of the biggest names in finance, communications and technology starting trials with VDO and Red Hat Ceph Storage.   A recently published IDC paper examines The Economics of  Software Defined Storage. In it, the…

Read more

azure

Unified Deduplication at SNIAs Software Development Conference

| Source
azure

Last week I attended the SNIA Software Development Conference (SDC) in Santa Clara, CA.  Some of the most interesting sessions included discussions about new erasure codes, new techniques for hardware accelerated data reduction, and several advancements developing storage stacks compatible with the latest technologies from Microsoft.  I led a session on Unified Deduplication at SDC. 

In a room packed to the brim with storage developers, my SDC talk focused on the differences between deduplication for backup and deduplication for primary storage.  Until recently, there weren’t any products that effectively addressed the needs of both markets.  In my presentation I discuss the core technologies behind deduplication, how workflow requirements differ between backup and primary use cases, and how to unify the two environments.  The gist of the presentation is that today’s backup software and special-purpose appliances are poorly suited to deliver deduplication for primary storage. Primary storage deduplication could be an outstanding solution for backup if it only had a mechanism for handling the data alignment patterns found in backup streams.  Because primary deduplication solutions, such as Permabit’s VDO device mapper target for Linux, generally handle fixed-sized chunk sizes (VDO uses 4 KB chunks), until now have not been unable to deliver deduplication benefits when dealing with unaligned streams of data.

With the introduction of the Optimizer file system capability in VDO6, unaligned streams of data can be deduplicated. Optimizer acts as a layered file system, sitting above the standard Linux file systems such as ext4 and xfs.  When it sees a familiar format, such as a tar or zip, it automatically segments the data on intelligent boundaries, padding it so that it is 4 KB aligned.  If the file type cannot be identified, a generic rolling hash algorithm is used to locate the boundaries and pad the output. The resulting data and associated 4 KB aligned metadata get written out into a file in the underlying file system.  When the 4 KB blocks hit the deduplication system below, the likelihood of locating matching blocks is increased improving the amount of data reduction. 

To demonstrate the benefits of Optimizer, I backed up my own personal 19.3 GB of Microsoft Office documents to a Linux system running VDO6 without Optimizer and saw 16% savings from compression. Then I backed up the same data to VDO6 system running Optimzer and saw impressive savings of 68%! 

 

When multiple full backup images are stored, the savings increase substantially.  A second full backup increases the savings to 84%, a third to 92%, a fourth to 96% and so on.  You can see the results summarized in the table below.  Even in environments with a high change rate, deduplication is a proven solution which yields 20:1 savings on backup in traditional business environments.

 

But the most compelling fact is that VDO6 doesn’t just solve the backup problem, it’s also able to address primary storage workloads as well.  The same modern server hardware can deliver up to  650,000 mixed random IOPS and backup rates of over 2 GB/s!  At those speeds, a single storage server could easily be addressing two workloads simultaneously, for example data warehousing and end-point protection (for desktops and mobile devices) could be run on the same systems.  That’s true flexibility for the  data center!

Read more